terraform/internal/command/test_test.go

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

5807 lines
160 KiB
Go
Raw Permalink Normal View History

2023-08-16 14:04:52 -04:00
// Copyright IBM Corp. 2014, 2026
// SPDX-License-Identifier: BUSL-1.1
package command
import (
"bytes"
"context"
"encoding/json"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
"errors"
"fmt"
"io"
"os"
"path"
"path/filepath"
"regexp"
"runtime"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
"sort"
"strings"
"testing"
"time"
"github.com/google/go-cmp/cmp"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
"github.com/google/go-cmp/cmp/cmpopts"
"github.com/hashicorp/cli"
"github.com/zclconf/go-cty/cty"
"github.com/hashicorp/terraform/internal/addrs"
testing_command "github.com/hashicorp/terraform/internal/command/testing"
"github.com/hashicorp/terraform/internal/command/views"
"github.com/hashicorp/terraform/internal/configs"
"github.com/hashicorp/terraform/internal/configs/configload"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
"github.com/hashicorp/terraform/internal/getproviders"
"github.com/hashicorp/terraform/internal/initwd"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
teststates "github.com/hashicorp/terraform/internal/moduletest/states"
"github.com/hashicorp/terraform/internal/providers"
"github.com/hashicorp/terraform/internal/registry"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
"github.com/hashicorp/terraform/internal/states/statemgr"
"github.com/hashicorp/terraform/internal/terminal"
"github.com/hashicorp/terraform/internal/terraform"
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
"github.com/hashicorp/terraform/internal/tfdiags"
)
moduletest: EvalContext lives outside of the "terraform" package The initial implementation of the module testing language included some elements deeply embedded inside "package terraform", which is primarily concerned with the traditional Terraform language, which we might now call the "module language" to distinguish it from the test language and the stacks language. That was a reasonable design decision due to the modules runtime needing to evaluate some expressions in a similar way to how the modules language does it, but unfortunately it caused the test language runtime to depend a little too closely on some current implementation details of how the modules runtime deals with expression evaluation, and those details are about to change as we teach Terraform to be able to deal with unknown values in new places, as part of resolving this issue: https://github.com/hashicorp/terraform/issues/30937 This commit therefore refactors the test runtime a little so that it doesn't intrude so deeply into the module runtime's business. Instead of directly extending the module runtime's expression evaluator, we'll instead call into it in a similar way to how the "terraform console" command does, by wrapping the evaluation data produced by the module runtime to extend it with the only two special things that the test runtime needs: references to the output values of previous runs, and references to test-only variables that aren't declared as part of the module under test. This effectively means that the test runtime can now evaluate a superset of the expressions that would be valid in the global scope of the module under test using only the public API of the modules runtime, with no testing-specific evaluation logic embedded inside the modules runtime. This new strategy is also somewhat similar to how the stacks runtime calls into the modules runtime to evaluate components, but the stacks runtime only actually cares about output values and so doesn't need a full evaluation scope. This means the result here is a little more symmetrical with the approach we took in stacks, which will hopefully allow experience working on one to translate better to the other.
2023-11-30 21:14:01 -05:00
func TestTest_Runs(t *testing.T) {
tcs := map[string]struct {
override string
args []string
envVars map[string]string
expectedOut []string
expectedErr []string
expectedResourceCount int
code int
initCode int
skip bool
description string
}{
"simple_pass": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"top-dir-only-test-files": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"top-dir-only-nested-test-files": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"simple_pass_nested": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"simple_pass_count": {
expectedOut: []string{"1 passed, 0 failed."},
args: []string{"-run-parallelism", "1"},
code: 0,
},
"simple_pass_nested_alternate": {
args: []string{"-test-directory", "other"},
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"simple_pass_very_nested": {
args: []string{"-test-directory", "tests/subdir"},
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"simple_pass_cmd_parallel": {
override: "simple_pass",
args: []string{"-parallelism", "1"},
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
description: "simple_pass with parallelism set to 1",
},
"simple_pass_very_nested_alternate": {
override: "simple_pass_very_nested",
args: []string{"-test-directory", "./tests/subdir"},
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"simple_pass_bad_test_directory": {
override: "simple_pass",
args: []string{"-test-directory", "../tests"},
expectedErr: []string{"Invalid testing directory"},
code: 1,
},
"simple_pass_bad_test_directory_abs": {
override: "simple_pass",
args: []string{"-test-directory", "/home/username/config/tests"},
expectedErr: []string{"Invalid testing directory"},
code: 1,
},
"pass_with_locals": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"pass_with_outputs": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"pass_with_variables": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"plan_then_apply": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"expect_failures_checks": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"expect_failures_inputs": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"expect_failures_resources": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"expect_failures_outputs": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"expect_failures_checks_verbose": {
override: "expect_failures_checks",
args: []string{"-verbose"},
expectedOut: []string{"1 passed, 0 failed.", "Warning: Check block assertion failed"},
code: 0,
},
"expect_failures_inputs_verbose": {
override: "expect_failures_inputs",
args: []string{"-verbose"},
expectedOut: []string{"1 passed, 0 failed."},
expectedErr: []string{"Error: Invalid value for variable"},
code: 0,
},
"expect_failures_resources_verbose": {
override: "expect_failures_resources",
args: []string{"-verbose"},
expectedOut: []string{"1 passed, 0 failed."},
expectedErr: []string{"Error: Resource postcondition failed"},
code: 0,
},
"expect_failures_outputs_verbose": {
override: "expect_failures_outputs",
args: []string{"-verbose"},
expectedOut: []string{"1 passed, 0 failed."},
expectedErr: []string{"Error: Module output value precondition failed"},
code: 0,
},
"multiple_files": {
expectedOut: []string{"2 passed, 0 failed"},
code: 0,
},
"multiple_files_with_filter": {
override: "multiple_files",
args: []string{"-filter=one.tftest.hcl"},
expectedOut: []string{"1 passed, 0 failed"},
code: 0,
},
"no_state": {
expectedOut: []string{"0 passed, 1 failed"},
expectedErr: []string{"No value for required variable"},
description: "the run apply fails, causing it to produce a nil state.",
code: 1,
},
"variables": {
expectedOut: []string{"2 passed, 0 failed"},
code: 0,
},
"variables_overridden": {
override: "variables",
args: []string{"-var=input=foo"},
expectedOut: []string{"1 passed, 1 failed"},
expectedErr: []string{`invalid value`},
code: 1,
},
"simple_fail": {
expectedOut: []string{"0 passed, 1 failed."},
expectedErr: []string{"invalid value", ` - "bar"
+ "zap"`},
code: 1,
},
"custom_condition_checks": {
expectedOut: []string{"0 passed, 1 failed."},
expectedErr: []string{"this really should fail"},
code: 1,
},
"custom_condition_inputs": {
expectedOut: []string{"0 passed, 1 failed."},
expectedErr: []string{"this should definitely fail"},
code: 1,
},
"custom_condition_outputs": {
expectedOut: []string{"0 passed, 1 failed."},
expectedErr: []string{"this should fail"},
code: 1,
},
"custom_condition_resources": {
expectedOut: []string{"0 passed, 1 failed."},
expectedErr: []string{"this really should fail"},
code: 1,
},
"no_providers_in_main": {
expectedOut: []string{"1 passed, 0 failed"},
code: 0,
},
"default_variables": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"undefined_variables": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"shared_state": {
expectedOut: []string{"8 passed, 0 failed."},
code: 0,
},
"shared_state_object": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"variable_references": {
expectedOut: []string{"3 passed, 0 failed."},
args: []string{"-var=global=\"triple\""},
code: 0,
},
"unreferenced_global_variable": {
override: "variable_references",
expectedOut: []string{"3 passed, 0 failed."},
// The other variable shouldn't pass validation, but it won't be
// referenced anywhere so should just be ignored.
args: []string{"-var=global=\"triple\"", "-var=other=bad"},
code: 0,
},
"variables_types": {
expectedOut: []string{"1 passed, 0 failed."},
args: []string{"-var=number_input=0", "-var=string_input=Hello, world!", "-var=list_input=[\"Hello\",\"world\"]"},
code: 0,
},
"null-outputs": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"destroy_fail": {
expectedOut: []string{"3 passed, 0 failed."},
expectedErr: []string{`Terraform left the following resources in state`},
code: 1,
expectedResourceCount: 4,
},
"default_optional_values": {
expectedOut: []string{"4 passed, 0 failed."},
code: 0,
},
"tfvars_in_test_dir": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"auto_tfvars_in_test_dir": {
override: "tfvars_in_test_dir",
args: []string{"-test-directory=alternate"},
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"functions_available": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"provider-functions-available": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"mocking": {
expectedOut: []string{"10 passed, 0 failed."},
code: 0,
},
"mocking-invalid": {
expectedErr: []string{
"The override_during attribute must be a value of plan or apply.",
},
initCode: 1,
},
"mocking-error": {
expectedErr: []string{
"Unknown condition value",
"plan_mocked_overridden.tftest.hcl",
"test_resource.primary[0].id",
"plan_mocked_provider.tftest.hcl",
"test_resource.secondary[0].id",
},
code: 1,
},
"dangling_data_block": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"skip_destroy_on_empty": {
expectedOut: []string{"3 passed, 0 failed."},
code: 0,
},
"empty_module_with_output": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"global_var_refs": {
expectedOut: []string{"1 passed, 0 failed, 2 skipped."},
expectedErr: []string{"The input variable \"env_var_input\" does not exist within this test file", "The input variable \"setup\" does not exist within this test file"},
code: 1,
},
"global_var_ref_in_suite_var": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"env-vars": {
expectedOut: []string{"1 passed, 0 failed."},
envVars: map[string]string{
"TF_VAR_input": "foo",
},
code: 0,
},
"env-vars-in-module": {
expectedOut: []string{"2 passed, 0 failed."},
envVars: map[string]string{
"TF_VAR_input": "foo",
},
code: 0,
},
"ephemeral_input": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"ephemeral_input_with_error": {
expectedOut: []string{"Error message refers to ephemeral values", "1 passed, 1 failed."},
expectedErr: []string{"Test assertion failed",
` - "(ephemeral value)"
+ "bar"`},
code: 1,
},
"ephemeral_resource": {
expectedOut: []string{"0 passed, 1 failed."},
2024-10-16 11:10:13 -04:00
expectedErr: []string{"Ephemeral resource instance has expired", "Ephemeral resources cannot be asserted"},
code: 1,
},
"with_state_key": {
expectedOut: []string{"3 passed, 1 failed."},
expectedErr: []string{"Test assertion failed", "resource renamed without moved block"},
code: 1,
},
"unapplyable-plan": {
expectedOut: []string{"0 passed, 1 failed."},
expectedErr: []string{"Cannot apply non-applyable plan"},
code: 1,
},
"write-only-attributes": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"write-only-attributes-mocked": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"write-only-attributes-overridden": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"with-default-variables": {
args: []string{"-var=input_two=universe"},
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
"parallel-errors": {
expectedOut: []string{"1 passed, 1 failed, 1 skipped."},
expectedErr: []string{"Invalid condition run"},
code: 1,
},
"write-into-default-state": {
args: []string{"-verbose"},
expectedOut: []string{"test_resource.two will be destroyed"},
code: 0,
},
"prevent-destroy": {
expectedOut: []string{"1 passed, 0 failed."},
code: 0,
},
"deferred_changes": {
args: []string{"-allow-deferral"},
expectedOut: []string{"3 passed, 0 failed."},
code: 0,
},
"ephemeral_output": {
code: 0,
},
"ephemeral_output_referenced": {
code: 0,
},
"no-tests": {
code: 0,
},
"simple_pass_function": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
expectedResourceCount: 0,
},
"mocking-invalid-outputs": {
expectedErr: []string{
"Invalid outputs attribute",
},
code: 1,
},
"mock-sources-inline": {
expectedOut: []string{"2 passed, 0 failed."},
code: 0,
},
}
for name, tc := range tcs {
t.Run(name, func(t *testing.T) {
if tc.skip {
t.Skip()
}
for k, v := range tc.envVars {
os.Setenv(k, v)
}
defer func() {
for k := range tc.envVars {
os.Unsetenv(k)
}
}()
file := name
if len(tc.override) > 0 {
file = tc.override
}
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", file)), td)
t.Chdir(td)
store := &testing_command.ResourceStore{
Data: make(map[string]cty.Value),
}
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: &testingOverrides{
Providers: map[addrs.Provider]providers.Factory{
addrs.NewDefaultProvider("test"): func() (providers.Interface, error) {
return testing_command.NewProvider(store).Provider, nil
},
},
},
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != tc.initCode {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", tc.initCode, code, output.All())
}
if tc.initCode > 0 {
// Then we don't expect the init step to succeed. So we'll check
// the init output for our expected error messages and outputs.
output := done(t).All()
stdout, stderr := output, output
if len(tc.expectedOut) > 0 {
for _, expectedOut := range tc.expectedOut {
if !strings.Contains(stdout, expectedOut) {
t.Errorf("output didn't contain expected string:\n\n%s", stdout)
}
}
}
if len(tc.expectedErr) > 0 {
for _, expectedErr := range tc.expectedErr {
if !strings.Contains(stderr, expectedErr) {
t.Errorf("error didn't contain expected string:\n\n%s", stderr)
}
}
} else if stderr != "" {
t.Errorf("unexpected stderr output\n%s", stderr)
}
// If `terraform init` failed, then we don't expect that
// `terraform test` will have run at all, so we can just return
// here.
return
}
// discard the output from the init command
done(t)
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run(append(tc.args, "-no-color"))
output := done(t)
if code != tc.code {
t.Errorf("expected status code %d but got %d:\n\n%s", tc.code, code, output.All())
}
if len(tc.expectedOut) > 0 {
for _, expectedOut := range tc.expectedOut {
if !strings.Contains(output.Stdout(), expectedOut) {
t.Errorf("output didn't contain expected string (%q):\n\n%s", expectedOut, output.Stdout())
}
}
}
if len(tc.expectedErr) > 0 {
for _, expectedErr := range tc.expectedErr {
if !strings.Contains(output.Stderr(), expectedErr) {
t.Errorf("error didn't contain expected string (%q):\n\n%s", expectedErr, output.Stderr())
}
}
} else if output.Stderr() != "" {
moduletest: EvalContext lives outside of the "terraform" package The initial implementation of the module testing language included some elements deeply embedded inside "package terraform", which is primarily concerned with the traditional Terraform language, which we might now call the "module language" to distinguish it from the test language and the stacks language. That was a reasonable design decision due to the modules runtime needing to evaluate some expressions in a similar way to how the modules language does it, but unfortunately it caused the test language runtime to depend a little too closely on some current implementation details of how the modules runtime deals with expression evaluation, and those details are about to change as we teach Terraform to be able to deal with unknown values in new places, as part of resolving this issue: https://github.com/hashicorp/terraform/issues/30937 This commit therefore refactors the test runtime a little so that it doesn't intrude so deeply into the module runtime's business. Instead of directly extending the module runtime's expression evaluator, we'll instead call into it in a similar way to how the "terraform console" command does, by wrapping the evaluation data produced by the module runtime to extend it with the only two special things that the test runtime needs: references to the output values of previous runs, and references to test-only variables that aren't declared as part of the module under test. This effectively means that the test runtime can now evaluate a superset of the expressions that would be valid in the global scope of the module under test using only the public API of the modules runtime, with no testing-specific evaluation logic embedded inside the modules runtime. This new strategy is also somewhat similar to how the stacks runtime calls into the modules runtime to evaluate components, but the stacks runtime only actually cares about output values and so doesn't need a full evaluation scope. This means the result here is a little more symmetrical with the approach we took in stacks, which will hopefully allow experience working on one to translate better to the other.
2023-11-30 21:14:01 -05:00
t.Errorf("unexpected stderr output\n%s", output.Stderr())
}
if len(store.Data) != tc.expectedResourceCount {
t.Errorf("should have left %d resources on completion but left %v", tc.expectedResourceCount, len(store.Data))
}
})
}
}
func TestTest_Interrupt(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "with_interrupt")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
interrupt := make(chan struct{})
provider.Interrupt = interrupt
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
},
}
c.Run(nil)
output := done(t).All()
if !strings.Contains(output, "Interrupt received") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
if provider.ResourceCount() > 0 {
// we asked for a nice stop in this one, so it should still have tidied everything up.
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_DestroyFail(t *testing.T) {
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// Testing that when a cleanup fails, we leave behind state files of the failed
// resources, and that the test command fails with a non-zero exit code.
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "destroy_fail")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
view, done := testView(t)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ProviderSource: providerSource,
}
init := &InitCommand{Meta: meta}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 0, code, output.All())
}
interrupt := make(chan struct{})
provider.Interrupt = interrupt
view, done = testView(t)
c := &TestCommand{
Meta: Meta{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
AllowExperimentalFeatures: true,
},
}
c.Run([]string{"-no-color"})
output := done(t)
err := output.Stderr()
cleanupMessage := `main.tftest.hcl... in progress
run "setup"... pass
run "single"... pass
run "double"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 3 passed, 0 failed.
`
cleanupErr := `Terraform encountered an error destroying resources created while executing
main.tftest.hcl/double.
Error: Failed to destroy resource
destroy_fail is set to true
Error: Failed to destroy resource
destroy_fail is set to true
Terraform left the following resources in state after executing
main.tftest.hcl/double, and they need to be cleaned up manually:
- test_resource.another
- test_resource.resource
Terraform encountered an error destroying resources created while executing
main.tftest.hcl/single.
Error: Failed to destroy resource
destroy_fail is set to true
Error: Failed to destroy resource
destroy_fail is set to true
Terraform left the following resources in state after executing
main.tftest.hcl/single, and they need to be cleaned up manually:
- test_resource.another
- test_resource.resource
`
// It's really important that the above message is printed, so we're testing
// for it specifically and making sure it contains all the resources.
if diff := cmp.Diff(cleanupErr, err); diff != "" {
t.Errorf("expected err to be\n%s\n\nbut got\n%s\n\n diff:\n%s\n", cleanupErr, err, diff)
}
if diff := cmp.Diff(cleanupMessage, output.Stdout()); diff != "" {
t.Errorf("expected output to be \n%s\n\nbut got \n%s\n\n diff:\n%s\n", cleanupMessage, output.Stdout(), diff)
}
if provider.ResourceCount() != 4 {
t.Errorf("should not have deleted all resources on completion but only has %v", provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expectedStates := map[string][]string{
"main.": {"test_resource.another", "test_resource.resource"},
"main.double": {"test_resource.another", "test_resource.resource"},
}
if diff := cmp.Diff(expectedStates, statesFromManifest(t, td)); diff != "" {
t.Fatalf("unexpected states: %s", diff)
}
t.Run("cleanup failed state", func(t *testing.T) {
interrupt := make(chan struct{})
provider.Interrupt = interrupt
provider.Provider.PlanResourceChangeFn = func(req providers.PlanResourceChangeRequest) providers.PlanResourceChangeResponse {
return providers.PlanResourceChangeResponse{
PlannedState: req.ProposedNewState,
}
}
provider.Provider.ApplyResourceChangeFn = func(req providers.ApplyResourceChangeRequest) providers.ApplyResourceChangeResponse {
return providers.ApplyResourceChangeResponse{
NewState: req.PlannedState,
}
}
view, done = testView(t)
c := &TestCleanupCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
AllowExperimentalFeatures: true,
},
}
c.Run([]string{"-no-color"})
output := done(t)
actualCleanup := output.Stdout()
expectedCleanup := `main.tftest.hcl... in progress
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success!
`
if diff := cmp.Diff(expectedCleanup, actualCleanup); diff != "" {
t.Fatalf("unexpected cleanup output: expected %s\n, got %s\n, diff: %s", expectedCleanup, actualCleanup, diff)
}
expectedStates := map[string][]string{}
if diff := cmp.Diff(expectedStates, statesFromManifest(t, td)); diff != "" {
t.Fatalf("unexpected states after cleanup: %s", diff)
}
})
}
func TestTest_Cleanup(t *testing.T) {
// function to consolidate the test command that should generate some state files and manifest
// It also does assertions.
executeTestCmd := func(provider *testing_command.TestProvider, providerSource *getproviders.MockSource) (td string) {
td = t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "cleanup")), td)
t.Chdir(td)
view, done := testView(t)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{Meta: meta}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 0, code, output.All())
}
interrupt := make(chan struct{})
provider.Interrupt = interrupt
view, done = testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
AllowExperimentalFeatures: true,
},
}
c.Run([]string{"-no-color"})
output := done(t)
message := `main.tftest.hcl... in progress
run "test"... pass
run "test_two"... pass
run "test_three"... pass
run "test_four"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 4 passed, 0 failed.
`
outputErr := `Terraform encountered an error destroying resources created while executing
main.tftest.hcl/test_three.
Error: Failed to destroy resource
destroy_fail is set to true
Terraform left the following resources in state after executing
main.tftest.hcl/test_three, and they need to be cleaned up manually:
- test_resource.resource
`
if diff := cmp.Diff(outputErr, output.Stderr()); diff != "" {
t.Errorf("expected err to be %s\n\nbut got %s\n\n diff:%s\n", outputErr, output.Stderr(), diff)
}
if diff := cmp.Diff(message, output.Stdout()); diff != "" {
t.Errorf("expected output to be %s\n\nbut got %s\n\n diff:%s\n", message, output.Stdout(), diff)
}
if provider.ResourceCount() != 2 {
t.Errorf("should have 2 resources on completion but only has %v", provider.ResourceString())
}
expectedStates := map[string][]string{
"main.": {"test_resource.resource"},
"main.state_three": {"test_resource.resource"},
}
actual := removeOutputs(statesFromManifest(t, td))
if diff := cmp.Diff(expectedStates, actual); diff != "" {
t.Fatalf("unexpected states: %s", diff)
}
return
}
t.Run("cleanup all left-over state", func(t *testing.T) {
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
// Run the test command to create the state
td := executeTestCmd(provider, providerSource)
interrupt := make(chan struct{})
provider.Interrupt = interrupt
provider.Provider.PlanResourceChangeFn = func(req providers.PlanResourceChangeRequest) providers.PlanResourceChangeResponse {
return providers.PlanResourceChangeResponse{
PlannedState: req.ProposedNewState,
}
}
provider.Provider.ApplyResourceChangeFn = func(req providers.ApplyResourceChangeRequest) providers.ApplyResourceChangeResponse {
var diags tfdiags.Diagnostics
// Simulate an error during apply, unless it is a destroy operation
if !req.PlannedState.IsNull() {
diags = diags.Append(fmt.Errorf("apply error"))
}
return providers.ApplyResourceChangeResponse{
NewState: req.PlannedState,
Diagnostics: diags,
}
}
view, done := testView(t)
c := &TestCleanupCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
AllowExperimentalFeatures: true,
},
}
c.Run([]string{"-no-color"})
output := done(t)
expectedCleanup := `main.tftest.hcl... in progress
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success!
`
if diff := cmp.Diff(expectedCleanup, output.Stdout()); diff != "" {
t.Errorf("unexpected cleanup output: expected %s\n, got %s\n, diff: %s", expectedCleanup, output.Stdout(), diff)
}
expectedStates := map[string][]string{}
actualStates := removeOutputs(statesFromManifest(t, td))
if diff := cmp.Diff(expectedStates, actualStates); diff != "" {
t.Fatalf("unexpected states after cleanup: %s", diff)
}
})
t.Run("cleanup failed state only (-repair)", func(t *testing.T) {
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
// Run the test command to create the state
td := executeTestCmd(provider, providerSource)
interrupt := make(chan struct{})
provider.Interrupt = interrupt
provider.Provider.PlanResourceChangeFn = func(req providers.PlanResourceChangeRequest) providers.PlanResourceChangeResponse {
return providers.PlanResourceChangeResponse{
PlannedState: req.ProposedNewState,
}
}
provider.Provider.ApplyResourceChangeFn = func(req providers.ApplyResourceChangeRequest) providers.ApplyResourceChangeResponse {
return providers.ApplyResourceChangeResponse{
NewState: req.PlannedState,
}
}
view, done := testView(t)
c := &TestCleanupCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
AllowExperimentalFeatures: true,
},
}
c.Run([]string{"-no-color", "-repair"})
output := done(t)
expectedCleanup := `main.tftest.hcl... in progress
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success!
`
if diff := cmp.Diff(expectedCleanup, output.Stdout()); diff != "" {
t.Fatalf("unexpected cleanup output: expected %s\n, got %s\n, diff: %s", expectedCleanup, output.Stdout(), diff)
}
expectedStates := map[string][]string{
"main.": {"test_resource.resource"},
}
actual := removeOutputs(statesFromManifest(t, td))
if diff := cmp.Diff(expectedStates, actual); diff != "" {
t.Fatalf("unexpected states after cleanup: %s", diff)
}
})
}
func TestTest_CleanupActuallyCleansUp(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "skip_cleanup_simple")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
// Run the cleanup command.
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
cleanup := &TestCleanupCommand{
Meta: meta,
}
code = cleanup.Run([]string{"-no-color"})
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
// Running the test again should now work, because we cleaned everything
// up.
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c = &TestCommand{
Meta: meta,
}
code = c.Run([]string{"-no-color"})
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
}
func TestTest_SkipCleanup_ConsecutiveTestsFail(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "skip_cleanup_simple")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
// Running the test again should fail because of the skip cleanup.
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c = &TestCommand{
Meta: meta,
}
code = c.Run([]string{"-no-color"})
output = done(t)
if code == 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expectedOut := "main.tftest.hcl... in progress\nmain.tftest.hcl... tearing down\nmain.tftest.hcl... fail\n\nFailure! 0 passed, 0 failed.\n"
expectedErr := "\nError: State manifest not empty\n\nThe state manifest for main.tftest.hcl should be empty before running tests.\nThis could be due to a previous test run not cleaning up after itself. Please\nensure that all state files are cleaned up before running tests.\n"
if diff := cmp.Diff(expectedOut, output.Stdout()); len(diff) > 0 {
t.Error(diff)
}
if diff := cmp.Diff(expectedErr, output.Stderr()); len(diff) > 0 {
t.Error(diff)
}
}
func TestTest_SharedState_Order(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "shared_state")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 9, code, output.All())
}
c := &TestCommand{
Meta: meta,
}
c.Run(nil)
output := done(t).All()
// Split the log into lines
lines := strings.Split(output, "\n")
var arr []string
for _, line := range lines {
if strings.Contains(line, "run \"") && strings.Contains(line, "\x1b[32mpass") {
arr = append(arr, line)
}
}
// Ensure the order of the tests is correct. Even though they share no state,
// the order should be sequential.
expectedOrder := []string{
// main.tftest.hcl
"run \"setup\"",
"run \"test\"",
// no-shared-state.tftest.hcl
"run \"setup\"",
"run \"test_a\"",
"run \"test_b\"",
"run \"test_c\"",
"run \"test_d\"",
"run \"test_e\"",
}
for i, line := range expectedOrder {
if !strings.Contains(arr[i], line) {
t.Errorf("unexpected test order: expected %q, got %q", line, arr[i])
}
}
}
func TestTest_Parallel_Divided_Order(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "parallel_divided")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 9, code, output.All())
}
c := &TestCommand{
Meta: meta,
}
c.Run(nil)
output := done(t).All()
// Split the log into lines
lines := strings.Split(output, "\n")
// Find the positions of the tests in the log output
var mainFirstIndex, mainSecondIndex, mainThirdIndex, mainFourthIndex, mainFifthIndex, mainSixthIndex int
for i, line := range lines {
if strings.Contains(line, "run \"main_first\"") {
mainFirstIndex = i
} else if strings.Contains(line, "run \"main_second\"") {
mainSecondIndex = i
} else if strings.Contains(line, "run \"main_third\"") {
mainThirdIndex = i
} else if strings.Contains(line, "run \"main_fourth\"") {
mainFourthIndex = i
} else if strings.Contains(line, "run \"main_fifth\"") {
mainFifthIndex = i
} else if strings.Contains(line, "run \"main_sixth\"") {
mainSixthIndex = i
}
}
if mainFirstIndex == 0 || mainSecondIndex == 0 || mainThirdIndex == 0 || mainFourthIndex == 0 || mainFifthIndex == 0 || mainSixthIndex == 0 {
t.Fatalf("one or more tests not found in the log output")
}
// Ensure the order of the tests is correct. The runs before main_fourth can execute in parallel.
if mainFirstIndex > mainFourthIndex || mainSecondIndex > mainFourthIndex || mainThirdIndex > mainFourthIndex {
t.Errorf("main_first, main_second, or main_third appears after main_fourth in the log output")
}
// Ensure main_fifth and main_sixth do not execute before main_fourth
if mainFifthIndex < mainFourthIndex {
t.Errorf("main_fifth appears before main_fourth in the log output")
}
if mainSixthIndex < mainFourthIndex {
t.Errorf("main_sixth appears before main_fourth in the log output")
}
}
func TestTest_Parallel(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "parallel")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 9, code, output.All())
}
c := &TestCommand{
Meta: meta,
}
c.Run(nil)
output := done(t).All()
if !strings.Contains(output, "40 passed, 0 failed") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
// Split the log into lines
lines := strings.Split(output, "\n")
// Find the positions of "test_d", "test_c", "test_setup" in the log output
var testDIndex, testCIndex, testSetupIndex int
for i, line := range lines {
if strings.Contains(line, "run \"setup\"") {
testSetupIndex = i
} else if strings.Contains(line, "run \"test_d\"") {
testDIndex = i
} else if strings.Contains(line, "run \"test_c\"") {
testCIndex = i
}
}
if testDIndex == 0 || testCIndex == 0 || testSetupIndex == 0 {
t.Fatalf("test_d, test_c, or test_setup not found in the log output")
}
// Ensure "test_d" appears before "test_c", because test_d has no dependencies,
// and would therefore run in parallel to much earlier tests which test_c depends on.
if testDIndex > testCIndex {
t.Errorf("test_d appears after test_c in the log output")
}
// Ensure "test_d" appears after "test_setup", because they have the same state key
if testDIndex < testSetupIndex {
t.Errorf("test_d appears before test_setup in the log output")
}
}
func TestTest_ParallelTeardown(t *testing.T) {
tests := []struct {
name string
sources map[string]string
assertFunc func(t *testing.T, output string, dur time.Duration)
}{
{
name: "parallel teardown",
sources: map[string]string{
"main.tf": `
variable "input" {
type = string
}
resource "test_resource" "foo" {
value = var.input
destroy_wait_seconds = 3
}
output "value" {
value = test_resource.foo.value
}
`,
"parallel.tftest.hcl": `
test {
parallel = true
}
variables {
foo = "foo"
}
provider "test" {
}
provider "test" {
alias = "start"
}
run "test_a" {
state_key = "state_foo"
variables {
input = "foo"
}
providers = {
test = test
}
assert {
condition = output.value == var.foo
error_message = "error in test_a"
}
}
run "test_b" {
state_key = "state_bar"
variables {
input = "bar"
}
providers = {
test = test.start
}
assert {
condition = output.value == "bar"
error_message = "error in test_b"
}
}
`,
},
assertFunc: func(t *testing.T, output string, dur time.Duration) {
if !strings.Contains(output, "2 passed, 0 failed") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
// Each teardown sleeps for 3 seconds, so we expect the total duration to be less than 6 seconds.
if dur >= 6*time.Second {
t.Fatalf("parallel.tftest.hcl duration took too long: %0.2f seconds", dur.Seconds())
}
},
},
{
name: "reference prevents parallel teardown",
sources: map[string]string{
"main.tf": `
variable "input" {
type = string
}
resource "test_resource" "foo" {
value = var.input
destroy_wait_seconds = 5
}
output "value" {
value = test_resource.foo.value
}
`,
"parallel.tftest.hcl": `
test {
parallel = true
}
variables {
foo = "foo"
}
provider "test" {
}
provider "test" {
alias = "start"
}
run "test_a" {
state_key = "state_foo"
variables {
input = "foo"
}
providers = {
test = test
}
assert {
condition = output.value == var.foo
error_message = "error in test_a"
}
}
run "test_b" {
state_key = "state_bar"
variables {
input = "bar"
}
providers = {
test = test.start
}
assert {
condition = output.value != run.test_a.value
error_message = "error in test_b"
}
}
`,
},
assertFunc: func(t *testing.T, output string, dur time.Duration) {
if !strings.Contains(output, "2 passed, 0 failed") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
// Each teardown sleeps for 5 seconds, so we expect the total duration to be at least 10 seconds.
if dur < 10*time.Second {
t.Fatalf("parallel.tftest.hcl duration took too short: %0.2f seconds", dur.Seconds())
}
},
},
{
name: "possible cyclic state key reference: skip edge that would cause cycle",
sources: map[string]string{
"main.tf": `
variable "foo" {
type = string
}
resource "test_resource" "foo" {
value = var.foo
// destroy_wait_seconds = 5
}
output "value" {
value = test_resource.foo.value
}
`,
// c2 => a1, b1 => a1, a2 => b1, b2 => c1
"parallel.tftest.hcl": `
test {
parallel = true
}
variables {
foo = "foo"
indirect = run.c1.value
}
provider "test" {
}
provider "test" {
alias = "start"
}
run "a1" {
state_key = "a"
variables {
foo = "foo"
}
providers = {
test = test
}
}
run "b1" {
state_key = "b"
variables {
foo = run.a1.value // no destroy edge here, because b2 owns the destroy node.
}
providers = {
test = test
}
}
run "a2" {
state_key = "a"
variables {
foo = run.b1.value
}
providers = {
test = test
}
}
run "b2" {
state_key = "b"
variables {
foo = var.indirect # This is an indirect reference to run.c1.value
unused = run.b1.value
}
providers = {
test = test
}
}
run "c1" {
state_key = "c"
variables {
foo = "foo"
}
}
run "c2" {
state_key = "c"
variables {
foo = run.a1.value
}
}
`,
},
assertFunc: func(t *testing.T, output string, dur time.Duration) {
if !strings.Contains(output, "6 passed, 0 failed") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
lines := strings.Split(output, "\n")
aIdx, bIdx, cIdx := -1, -1, -1
for idx, line := range lines {
if strings.Contains(line, "tearing down") {
if strings.Contains(line, "a2") {
aIdx = idx
}
if strings.Contains(line, "b2") {
bIdx = idx
}
if strings.Contains(line, "c2") {
cIdx = idx
}
}
}
if cIdx > aIdx || aIdx > bIdx { // c => a => b
t.Errorf("teardown order is incorrect: c2 (%d), a2 (%d), b2 (%d)", cIdx, aIdx, bIdx)
}
},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
_, td, closer := testModuleInline(t, tt.sources)
defer closer()
t.Chdir(td)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
// create a new provider instance for each test run, so that we can
// ensure that the test provider locks do not interfere between runs.
pInst := func() providers.Interface {
return testing_command.NewProvider(nil).Provider
}
meta := Meta{
testingOverrides: &testingOverrides{
Providers: map[addrs.Provider]providers.Factory{
addrs.NewDefaultProvider("test"): func() (providers.Interface, error) {
return pInst(), nil
},
}},
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{Meta: meta}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 0, code, output.All())
}
c := &TestCommand{Meta: meta}
c.Run([]string{"-json", "-no-color"})
output := done(t).All()
// Split the log into lines
lines := strings.Split(output, "\n")
// Find the start of the teardown and complete timestamps
var startTimestamp, completeTimestamp string
for _, line := range lines {
if strings.Contains(line, `{"path":"parallel.tftest.hcl","progress":"teardown"`) {
var obj map[string]interface{}
if err := json.Unmarshal([]byte(line), &obj); err == nil {
if ts, ok := obj["@timestamp"].(string); ok {
startTimestamp = ts
}
}
} else if strings.Contains(line, `{"path":"parallel.tftest.hcl","progress":"complete"`) {
var obj map[string]interface{}
if err := json.Unmarshal([]byte(line), &obj); err == nil {
if ts, ok := obj["@timestamp"].(string); ok {
completeTimestamp = ts
}
}
}
}
if startTimestamp == "" || completeTimestamp == "" {
t.Fatalf("could not find start or complete timestamp in log output")
}
startTime, err := time.Parse(time.RFC3339Nano, startTimestamp)
if err != nil {
t.Fatalf("failed to parse start timestamp: %v", err)
}
completeTime, err := time.Parse(time.RFC3339Nano, completeTimestamp)
if err != nil {
t.Fatalf("failed to parse complete timestamp: %v", err)
}
dur := completeTime.Sub(startTime)
if tt.assertFunc != nil {
tt.assertFunc(t, output, dur)
}
})
}
}
func TestTest_InterruptSkipsRemaining(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "with_interrupt_and_additional_file")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
interrupt := make(chan struct{})
provider.Interrupt = interrupt
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
},
}
c.Run([]string{"-no-color"})
output := done(t).All()
if !strings.Contains(output, "skip_me.tftest.hcl... skip") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
if provider.ResourceCount() > 0 {
// we asked for a nice stop in this one, so it should still have tidied everything up.
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_DoubleInterrupt(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "with_double_interrupt")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
interrupt := make(chan struct{})
provider.Interrupt = interrupt
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
ShutdownCh: interrupt,
},
}
c.Run(nil)
output := done(t).All()
if !strings.Contains(output, "Two interrupts received") {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
cleanupMessage := `Terraform was interrupted while executing main.tftest.hcl, and may not have
performed the expected cleanup operations.
Terraform has already created the following resources from the module under
test:
- test_resource.primary
- test_resource.secondary
- test_resource.tertiary`
// It's really important that the above message is printed, so we're testing
// for it specifically and making sure it contains all the resources.
if !strings.Contains(output, cleanupMessage) {
t.Errorf("output didn't produce the right output:\n\n%s", output)
}
// This time the test command shouldn't have cleaned up the resource because
// of the hard interrupt.
if provider.ResourceCount() != 3 {
// we asked for a nice stop in this one, so it should still have tidied everything up.
t.Errorf("should not have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_ProviderAlias(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "with_provider_alias")), td)
t.Chdir(td)
store := &testing_command.ResourceStore{
Data: make(map[string]cty.Value),
}
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: &testingOverrides{
Providers: map[addrs.Provider]providers.Factory{
addrs.NewDefaultProvider("test"): func() (providers.Interface, error) {
return testing_command.NewProvider(store).Provider, nil
},
},
},
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run([]string{"-no-color"}); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
command := &TestCommand{
Meta: meta,
}
code := command.Run([]string{"-no-color"})
output = done(t)
printedOutput := false
if code != 0 {
printedOutput = true
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
if len(store.Data) > 0 {
if !printedOutput {
t.Errorf("should have deleted all resources on completion but left %d\n\n%s", len(store.Data), output.All())
} else {
t.Errorf("should have deleted all resources on completion but left %d", len(store.Data))
}
}
}
func TestTest_ComplexCondition(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "complex_condition")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{"test": {"1.0.0"}})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run([]string{"-no-color"}); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
command := &TestCommand{
Meta: meta,
}
code := command.Run([]string{"-no-color"})
output = done(t)
printedOutput := false
if code != 1 {
printedOutput = true
t.Errorf("expected status code 1 but got %d: %s", code, output.All())
}
expectedOut := `main.tftest.hcl... in progress
run "validate_diff_types"... fail
run "validate_output"... fail
run "validate_complex_output"... fail
run "validate_complex_output_sensitive"... fail
run "validate_complex_output_pass"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 1 passed, 4 failed.
`
expectedErr := `
Error: Test assertion failed
on main.tftest.hcl line 37, in run "validate_diff_types":
37: condition = var.tr1 == var.tr2
Warning: LHS and RHS values are of different types
expected to fail
Error: Test assertion failed
on main.tftest.hcl line 44, in run "validate_output":
44: condition = output.foo == var.foo
Diff:
--- actual
+++ expected
{
- "bar": "notbaz",
+ "bar": "baz",
"matches": "matches",
- "qux": "quux",
- "xuq": "xuq"
+ "qux": "qux",
+ "xuq": "nope"
}
expected to fail due to different values
Error: Test assertion failed
on main.tftest.hcl line 52, in run "validate_complex_output":
52: condition = output.complex == var.bar
Warning: LHS and RHS values are of different types
Diff:
--- actual
+++ expected
{
"root": [
{
"bar": [
1
],
- "qux": "quux"
+ "qux": "qux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
expected to fail
Error: Test assertion failed
on main.tftest.hcl line 60, in run "validate_complex_output_sensitive":
60: condition = output.complex == output.complex_sensitive
Diff:
--- actual
+++ expected
- {
- "root": [
- {
- "bar": [
- 1
- ],
- "qux": "quux"
- },
- {
- "bar": [
- 2
- ],
- "qux": "quux"
- }
- ]
- }
+ "(sensitive value)"
expected to fail
`
if diff := cmp.Diff(output.Stdout(), expectedOut); len(diff) > 0 {
t.Errorf("\nexpected: \n%s\ngot: %s\ndiff: %s", expectedOut, output.All(), diff)
}
if diff := cmp.Diff(output.Stderr(), expectedErr); len(diff) > 0 {
t.Errorf("\nexpected stderr: \n%s\ngot: %s\ndiff: %s", expectedErr, output.Stderr(), diff)
}
if provider.ResourceCount() > 0 {
if !printedOutput {
t.Errorf("should have deleted all resources on completion but left %s\n\n%s", provider.ResourceString(), output.All())
} else {
t.Errorf("should have deleted all resources on completion but left %s", provider.ResourceString())
}
}
}
func TestTest_ComplexConditionVerbose(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "complex_condition")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{"test": {"1.0.0"}})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run([]string{"-no-color"}); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
command := &TestCommand{
Meta: meta,
}
code := command.Run([]string{"-no-color", "-verbose"})
output = done(t)
printedOutput := false
if code != 1 {
printedOutput = true
t.Errorf("expected status code 1 but got %d: %s", code, output.All())
}
expectedErr := `
Error: Test assertion failed
on main.tftest.hcl line 37, in run "validate_diff_types":
37: condition = var.tr1 == var.tr2
LHS:
{
"iops": null,
"size": 60
}
RHS:
{
"iops": null,
"size": 60
}
Warning: LHS and RHS values are of different types
var.tr1 is {
"iops": null,
"size": 60
}
var.tr2 is {
"iops": null,
"size": 60
}
expected to fail
Error: Test assertion failed
on main.tftest.hcl line 44, in run "validate_output":
44: condition = output.foo == var.foo
LHS:
{
"bar": "notbaz",
"matches": "matches",
"qux": "quux",
"xuq": "xuq"
}
RHS:
{
"bar": "baz",
"matches": "matches",
"qux": "qux",
"xuq": "nope"
}
Diff:
--- actual
+++ expected
{
- "bar": "notbaz",
+ "bar": "baz",
"matches": "matches",
- "qux": "quux",
- "xuq": "xuq"
+ "qux": "qux",
+ "xuq": "nope"
}
output.foo is {
"bar": "notbaz",
"matches": "matches",
"qux": "quux",
"xuq": "xuq"
}
var.foo is {
"bar": "baz",
"matches": "matches",
"qux": "qux",
"xuq": "nope"
}
expected to fail due to different values
Error: Test assertion failed
on main.tftest.hcl line 52, in run "validate_complex_output":
52: condition = output.complex == var.bar
LHS:
{
"root": [
{
"bar": [
1
],
"qux": "quux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
RHS:
{
"root": [
{
"bar": [
1
],
"qux": "qux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
Warning: LHS and RHS values are of different types
Diff:
--- actual
+++ expected
{
"root": [
{
"bar": [
1
],
- "qux": "quux"
+ "qux": "qux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
output.complex is {
"root": [
{
"bar": [
1
],
"qux": "quux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
var.bar is {
"root": [
{
"bar": [
1
],
"qux": "qux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
expected to fail
Error: Test assertion failed
on main.tftest.hcl line 60, in run "validate_complex_output_sensitive":
60: condition = output.complex == output.complex_sensitive
LHS:
{
"root": [
{
"bar": [
1
],
"qux": "quux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
RHS:
"(sensitive value)"
Diff:
--- actual
+++ expected
- {
- "root": [
- {
- "bar": [
- 1
- ],
- "qux": "quux"
- },
- {
- "bar": [
- 2
- ],
- "qux": "quux"
- }
- ]
- }
+ "(sensitive value)"
output.complex is {
"root": [
{
"bar": [
1
],
"qux": "quux"
},
{
"bar": [
2
],
"qux": "quux"
}
]
}
output.complex_sensitive is "(sensitive value)"
expected to fail
`
outputs := []string{
"main.tftest.hcl... in progress",
" run \"validate_diff_types\"... fail",
" run \"validate_output\"... fail",
" run \"validate_complex_output\"... fail",
" run \"validate_complex_output_sensitive\"... fail",
" run \"validate_complex_output_pass\"... pass",
"main.tftest.hcl... tearing down",
"main.tftest.hcl... fail",
"Failure! 1 passed, 4 failed.",
}
stdout := output.Stdout()
for _, expected := range outputs {
if !strings.Contains(stdout, expected) {
t.Errorf("output didn't contain expected output %q", expected)
}
}
if diff := cmp.Diff(output.Stderr(), expectedErr); len(diff) > 0 {
t.Errorf("\nexpected stderr: \n%s\ngot: %s\ndiff: %s", expectedErr, output.Stderr(), diff)
}
if provider.ResourceCount() > 0 {
if !printedOutput {
t.Errorf("should have deleted all resources on completion but left %s\n\n%s", provider.ResourceString(), output.All())
} else {
t.Errorf("should have deleted all resources on completion but left %s", provider.ResourceString())
}
}
}
func TestTest_ModuleDependencies(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "with_setup_module")), td)
t.Chdir(td)
// Our two providers will share a common set of values to make things
// easier.
store := &testing_command.ResourceStore{
Data: make(map[string]cty.Value),
}
// We set it up so the module provider will update the data sources
// available to the core mock provider.
test := testing_command.NewProvider(store)
setup := testing_command.NewProvider(store)
test.SetDataPrefix("data")
test.SetResourcePrefix("resource")
// Let's make the setup provider write into the data for test provider.
setup.SetResourcePrefix("data")
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
"setup": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: &testingOverrides{
Providers: map[addrs.Provider]providers.Factory{
addrs.NewDefaultProvider("test"): providers.FactoryFixed(test.Provider),
addrs.NewDefaultProvider("setup"): providers.FactoryFixed(setup.Provider),
},
},
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
command := &TestCommand{
Meta: meta,
}
code := command.Run(nil)
output := done(t)
printedOutput := false
if code != 0 {
printedOutput = true
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
if test.ResourceCount() > 0 {
if !printedOutput {
printedOutput = true
t.Errorf("should have deleted all resources on completion but left %s\n\n%s", test.ResourceString(), output.All())
} else {
t.Errorf("should have deleted all resources on completion but left %s", test.ResourceString())
}
}
if setup.ResourceCount() > 0 {
if !printedOutput {
t.Errorf("should have deleted all resources on completion but left %s\n\n%s", setup.ResourceString(), output.All())
} else {
t.Errorf("should have deleted all resources on completion but left %s", setup.ResourceString())
}
}
}
func TestTest_CatchesErrorsBeforeDestroy(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "invalid_default_state")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 1 {
t.Errorf("expected status code 0 but got %d", code)
}
expectedOut := `main.tftest.hcl... in progress
run "test"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`
expectedErr := `
Error: No value for required variable
on main.tf line 2:
2: variable "input" {
The module under test for run block "test" has a required variable "input"
with no set value. Use a -var or -var-file command line argument or add this
variable into a "variables" block within the test file or run block.
`
actualOut := output.Stdout()
actualErr := output.Stderr()
if diff := cmp.Diff(actualOut, expectedOut); len(diff) > 0 {
t.Errorf("std out didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("std err didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_Verbose(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "plan_then_apply")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-verbose", "-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "validate_test_resource"... pass
Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
+ create
Terraform will perform the following actions:
# test_resource.foo will be created
+ resource "test_resource" "foo" {
+ destroy_fail = (known after apply)
+ id = "constant_value"
+ value = "bar"
+ write_only = (write-only attribute)
}
Plan: 1 to add, 0 to change, 0 to destroy.
run "apply_test_resource"... pass
# test_resource.foo:
resource "test_resource" "foo" {
destroy_fail = false
id = "constant_value"
value = "bar"
write_only = (write-only attribute)
}
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 2 passed, 0 failed.
`
actual := output.All()
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_ValidatesBeforeExecution(t *testing.T) {
tcs := map[string]struct {
expectedOut string
expectedErr string
}{
"invalid": {
expectedOut: `main.tftest.hcl... in progress
run "invalid"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`,
expectedErr: `
Error: Invalid ` + "`expect_failures`" + ` reference
on main.tftest.hcl line 5, in run "invalid":
5: local.my_value,
You cannot expect failures from local.my_value. You can only expect failures
from checkable objects such as input variables, output values, check blocks,
managed resources and data sources.
`,
},
"invalid-module": {
expectedOut: `main.tftest.hcl... in progress
run "invalid"... fail
run "test"... skip
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed, 1 skipped.
`,
expectedErr: `
Error: Reference to undeclared input variable
on setup/main.tf line 3, in resource "test_resource" "setup":
3: value = var.not_real // Oh no!
An input variable with the name "not_real" has not been declared. This
variable can be declared with a variable "not_real" {} block.
`,
},
"missing-provider": {
expectedOut: `main.tftest.hcl... in progress
run "passes_validation"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`,
expectedErr: `
Error: Provider configuration not present
To work with test_resource.secondary its original provider configuration at
provider["registry.terraform.io/hashicorp/test"].secondary is required, but
it has been removed. This occurs when a provider configuration is removed
while objects created by that provider still exist in the state. Re-add the
provider configuration to destroy test_resource.secondary, after which you
can remove the provider configuration again.
`,
},
"missing-provider-in-run-block": {
expectedOut: `main.tftest.hcl... in progress
run "passes_validation"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`,
expectedErr: `
Error: Provider configuration not present
To work with test_resource.secondary its original provider configuration at
provider["registry.terraform.io/hashicorp/test"].secondary is required, but
it has been removed. This occurs when a provider configuration is removed
while objects created by that provider still exist in the state. Re-add the
provider configuration to destroy test_resource.secondary, after which you
can remove the provider configuration again.
`,
},
"missing-provider-definition-in-file": {
expectedOut: `main.tftest.hcl... in progress
run "passes_validation"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`,
expectedErr: `
Error: Missing provider definition for test
on main.tftest.hcl line 12, in run "passes_validation":
12: test = test
This provider block references a provider definition that does not exist.
`,
},
"missing-provider-in-test-module": {
expectedOut: `main.tftest.hcl... in progress
run "passes_validation_primary"... pass
run "passes_validation_secondary"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 1 passed, 1 failed.
`,
expectedErr: `
Error: Provider configuration not present
To work with test_resource.secondary its original provider configuration at
provider["registry.terraform.io/hashicorp/test"].secondary is required, but
it has been removed. This occurs when a provider configuration is removed
while objects created by that provider still exist in the state. Re-add the
provider configuration to destroy test_resource.secondary, after which you
can remove the provider configuration again.
`,
},
}
for file, tc := range tcs {
t.Run(file, func(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", file)), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 1 {
t.Errorf("expected status code 1 but got %d", code)
}
actualOut, expectedOut := output.Stdout(), tc.expectedOut
actualErr, expectedErr := output.Stderr(), tc.expectedErr
if !strings.Contains(actualOut, expectedOut) {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s", expectedOut, actualOut)
}
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("error didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
})
}
}
func TestTest_NestedSetupModules(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "with_nested_setup_modules")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, ui.ErrorWriter)
}
command := &TestCommand{
Meta: meta,
}
code := command.Run(nil)
output := done(t)
printedOutput := false
if code != 0 {
printedOutput = true
t.Errorf("expected status code 0 but got %d: %s", code, output.All())
}
if provider.ResourceCount() > 0 {
if !printedOutput {
t.Errorf("should have deleted all resources on completion but left %s\n\n%s", provider.ResourceString(), output.All())
} else {
t.Errorf("should have deleted all resources on completion but left %s", provider.ResourceString())
}
}
}
func TestTest_StatePropagation(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "state_propagation")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-verbose", "-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "initial_apply_example"... pass
# test_resource.module_resource:
resource "test_resource" "module_resource" {
destroy_fail = false
id = "df6h8as9"
value = "start"
write_only = (write-only attribute)
}
run "initial_apply"... pass
# test_resource.resource:
resource "test_resource" "resource" {
destroy_fail = false
id = "598318e0"
value = "start"
write_only = (write-only attribute)
}
run "plan_second_example"... pass
Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
+ create
Terraform will perform the following actions:
# test_resource.second_module_resource will be created
+ resource "test_resource" "second_module_resource" {
+ destroy_fail = (known after apply)
+ id = "b6a1d8cb"
+ value = "start"
+ write_only = (write-only attribute)
}
Plan: 1 to add, 0 to change, 0 to destroy.
run "plan_update"... pass
Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
~ update in-place
Terraform will perform the following actions:
# test_resource.resource will be updated in-place
~ resource "test_resource" "resource" {
id = "598318e0"
~ value = "start" -> "update"
# (2 unchanged attributes hidden)
}
Plan: 0 to add, 1 to change, 0 to destroy.
run "plan_update_example"... pass
Terraform used the selected providers to generate the following execution
plan. Resource actions are indicated with the following symbols:
~ update in-place
Terraform will perform the following actions:
# test_resource.module_resource will be updated in-place
~ resource "test_resource" "module_resource" {
id = "df6h8as9"
~ value = "start" -> "update"
# (2 unchanged attributes hidden)
}
Plan: 0 to add, 1 to change, 0 to destroy.
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 5 passed, 0 failed.
`
actual := output.All()
if diff := cmp.Diff(expected, actual); diff != "" {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_SkipCleanup(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "skip_cleanup")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
t.Run("skipped resources should not be deleted", func(t *testing.T) {
expected := `
Warning: Duplicate "skip_cleanup" block
on main.tftest.hcl line 15, in run "test_three":
15: skip_cleanup = true
The run "test_three" has a skip_cleanup attribute set, but shares state with
an earlier run "test_two" that also has skip_cleanup set. The later run takes
precedence, and this attribute is ignored for the earlier run.
main.tftest.hcl... in progress
run "test"... pass
run "test_two"... pass
run "test_three"... pass
run "test_four"... pass
run "test_five"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Success! 5 passed, 0 failed.
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
actual := output.All()
if !strings.Contains(actual, expected) {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, cmp.Diff(expected, actual))
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if provider.ResourceCount() != 1 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
val := provider.Store.Get(provider.ResourceString())
if val.GetAttr("value").AsString() != "test_three" {
t.Errorf("expected resource to have value 'test_three' but got %s", val.GetAttr("value").AsString())
}
})
t.Run("state should be persisted", func(t *testing.T) {
expectedStates := map[string][]string{
"main.": {"test_resource.resource"},
}
actualStates := removeOutputs(statesFromManifest(t, td))
if diff := cmp.Diff(expectedStates, actualStates); diff != "" {
t.Fatalf("unexpected states: %s", diff)
}
})
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_SkipCleanupWithRunDependencies(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "skip_cleanup_with_run_deps")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Meta: meta,
}
code := c.Run([]string{"-no-color"})
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
t.Run("skipped resources should not be deleted", func(t *testing.T) {
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expected := `main.tftest.hcl... in progress
run "test"... pass
run "test_two"... pass
run "test_three"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Success! 3 passed, 0 failed.
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
actual := output.All()
if !strings.Contains(actual, expected) {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, cmp.Diff(expected, actual))
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if provider.ResourceCount() != 1 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
val := provider.Store.Get(provider.ResourceString())
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if val.GetAttr("value").AsString() != "test_two" {
t.Errorf("expected resource to have value 'test_two' but got %s", val.GetAttr("value").AsString())
}
})
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// we want to check that we leave behind the state that was skipped
// and the states that it depends on
t.Run("state should be persisted", func(t *testing.T) {
expectedStates := map[string][]string{
"main.": {"output.id", "output.unused"},
"main.state": {"test_resource.resource", "output.id", "output.unused"},
}
actualStates := statesFromManifest(t, td)
if diff := cmp.Diff(expectedStates, actualStates, equalIgnoreOrder()); diff != "" {
t.Fatalf("unexpected states: %s", diff)
}
})
t.Run("cleanup all left-over state", func(t *testing.T) {
view, done := testView(t)
c := &TestCleanupCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
Ui: ui,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
},
}
c.Run([]string{"-no-color"})
output := done(t)
expectedCleanup := `main.tftest.hcl... in progress
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Success!
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if diff := cmp.Diff(expectedCleanup, output.Stdout()); diff != "" {
t.Errorf("unexpected cleanup output: expected %s\n, got %s\n, diff: %s", expectedCleanup, output.Stdout(), diff)
}
if err := output.Stderr(); len(err) != 0 {
t.Errorf("unexpected error: %s", err)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expectedStates := map[string][]string{}
actualStates := removeOutputs(statesFromManifest(t, td))
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if diff := cmp.Diff(expectedStates, actualStates); diff != "" {
t.Fatalf("unexpected states after cleanup: %s", diff)
}
})
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_SkipCleanup_JSON(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "skip_cleanup")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
code := c.Run([]string{"-no-color", "-json"})
output = done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
var messages []string
for ix, line := range strings.Split(output.All(), "\n") {
if len(line) == 0 {
// Skip empty lines.
continue
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if ix == 0 {
// skip the first one, it's version information
continue
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
var obj map[string]interface{}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if err := json.Unmarshal([]byte(line), &obj); err != nil {
t.Errorf("failed to unmarshal returned line: %s", line)
continue
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// Remove the timestamp as it changes every time.
delete(obj, "@timestamp")
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if obj["type"].(string) == "test_run" {
// Then we need to delete the `elapsed` field from within the run
// as it'll cause flaky tests.
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
run := obj["test_run"].(map[string]interface{})
if run["progress"].(string) != "complete" {
delete(run, "elapsed")
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
message, err := json.Marshal(obj)
if err != nil {
t.Errorf("failed to remarshal returned line: %s", line)
continue
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
messages = append(messages, string(message))
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
t.Run("skipped resources should not be deleted", func(t *testing.T) {
expected := []string{
`{"@level":"warn","@message":"Warning: Duplicate \"skip_cleanup\" block","@module":"terraform.ui","diagnostic":{"detail":"The run \"test_three\" has a skip_cleanup attribute set, but shares state with an earlier run \"test_two\" that also has skip_cleanup set. The later run takes precedence, and this attribute is ignored for the earlier run.","range":{"end":{"byte":163,"column":15,"line":15},"filename":"main.tftest.hcl","start":{"byte":151,"column":3,"line":15}},"severity":"warning","snippet":{"code":" skip_cleanup = true","context":"run \"test_three\"","highlight_end_offset":14,"highlight_start_offset":2,"start_line":15,"values":[]},"summary":"Duplicate \"skip_cleanup\" block"},"type":"diagnostic"}`,
`{"@level":"info","@message":"Found 1 file and 5 run blocks","@module":"terraform.ui","test_abstract":{"main.tftest.hcl":["test","test_two","test_three","test_four","test_five"]},"type":"test_abstract"}`,
`{"@level":"info","@message":"main.tftest.hcl... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","test_file":{"path":"main.tftest.hcl","progress":"starting"},"type":"test_file"}`,
`{"@level":"info","@message":" \"test\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"starting","run":"test"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test\"... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"complete","run":"test","status":"pass"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_two\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_two","test_run":{"path":"main.tftest.hcl","progress":"starting","run":"test_two"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_two\"... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_two","test_run":{"path":"main.tftest.hcl","progress":"complete","run":"test_two","status":"pass"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_three\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_three","test_run":{"path":"main.tftest.hcl","progress":"starting","run":"test_three"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_three\"... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_three","test_run":{"path":"main.tftest.hcl","progress":"complete","run":"test_three","status":"pass"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_four\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_four","test_run":{"path":"main.tftest.hcl","progress":"starting","run":"test_four"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_four\"... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_four","test_run":{"path":"main.tftest.hcl","progress":"complete","run":"test_four","status":"pass"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_five\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_five","test_run":{"path":"main.tftest.hcl","progress":"starting","run":"test_five"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test_five\"... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_five","test_run":{"path":"main.tftest.hcl","progress":"complete","run":"test_five","status":"pass"},"type":"test_run"}`,
`{"@level":"info","@message":"main.tftest.hcl... tearing down","@module":"terraform.ui","@testfile":"main.tftest.hcl","test_file":{"path":"main.tftest.hcl","progress":"teardown"},"type":"test_file"}`,
`{"@level":"info","@message":" \"test_three\"... tearing down","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test_three","test_run":{"path":"main.tftest.hcl","progress":"teardown","run":"test_three"},"type":"test_run"}`,
`{"@level":"info","@message":"main.tftest.hcl... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","test_file":{"path":"main.tftest.hcl","progress":"complete","status":"pass"},"type":"test_file"}`,
`{"@level":"info","@message":"Success! 5 passed, 0 failed.","@module":"terraform.ui","test_summary":{"errored":0,"failed":0,"passed":5,"skipped":0,"status":"pass"},"type":"test_summary"}`,
}
trimmedActual := strings.Join(messages, "\n")
trimmedExpected := strings.Join(expected, "\n")
if diff := cmp.Diff(trimmedExpected, trimmedActual); diff != "" {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", trimmedExpected, trimmedActual, diff)
}
if provider.ResourceCount() != 1 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
val := provider.Store.Get(provider.ResourceString())
if val.GetAttr("value").AsString() != "test_three" {
t.Errorf("expected resource to have value 'test_three' but got %s", val.GetAttr("value").AsString())
}
})
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_SkipCleanup_FileLevelFlag(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "skip_file_cleanup")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Meta: meta,
}
code := c.Run([]string{"-no-color"})
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output = done(t)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
t.Run("skipped resources should not be deleted", func(t *testing.T) {
expected := `main.tftest.hcl... in progress
run "test"... pass
run "test_two"... pass
run "test_three"... pass
run "test_four"... pass
run "test_five"... pass
main.tftest.hcl... tearing down
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
main.tftest.hcl... pass
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Success! 5 passed, 0 failed.
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
actual := output.All()
if !strings.Contains(actual, expected) {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s", expected, actual)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if provider.ResourceCount() != 1 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
val := provider.Store.Get(provider.ResourceString())
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if val.GetAttr("value").AsString() != "test_four" {
t.Errorf("expected resource to have value 'test_four' but got %s", val.GetAttr("value").AsString())
}
})
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
t.Run("state should be persisted with valid reason", func(t *testing.T) {
manifest, err := teststates.LoadManifest(td, true)
if err != nil {
t.Fatal(err)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expectedStates := map[string][]string{
"main.": {"test_resource.resource"},
}
actualStates := make(map[string][]string)
var reason teststates.StateReason
// Verify the states in the manifest
for fileName, file := range manifest.Files {
for name, state := range file.States {
sm := statemgr.NewFilesystem(manifest.StateFilePath(state.ID))
if err := sm.RefreshState(); err != nil {
t.Fatalf("error when reading state file: %s", err)
}
reason = state.Reason
state := sm.State()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// If the state is nil, then the test cleaned up the state
if state == nil {
t.Fatalf("state is nil")
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
var resources []string
for _, module := range state.Modules {
for _, resource := range module.Resources {
resources = append(resources, resource.Addr.String())
}
}
sort.Strings(resources)
actualStates[strings.TrimSuffix(fileName, ".tftest.hcl")+"."+name] = resources
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if diff := cmp.Diff(expectedStates, actualStates); diff != "" {
t.Fatalf("unexpected states: %s", diff)
}
if reason != teststates.StateReasonSkip {
t.Fatalf("expected reason to be skip but got %s", reason)
}
})
}
func TestTest_OnlyExternalModules(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "only_modules")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "first"... pass
run "second"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 2 passed, 0 failed.
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
actual := output.Stdout()
if !strings.Contains(actual, expected) {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s", expected, actual)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_PartialUpdates(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "partial_updates")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expected := `main.tftest.hcl... in progress
run "first"... pass
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Warning: Resource targeting is in effect
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
You are creating a plan with the -target option, which means that the result
of this plan may not represent all of the changes requested by the current
configuration.
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
The -target option is not for routine use, and is provided only for
exceptional situations such as recovering from errors or mistakes, or when
Terraform specifically suggests to use it as part of an error message.
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Warning: Applied changes may be incomplete
The plan was created with the -target option in effect, so some changes
requested in the configuration may have been ignored and the output values
may not be fully updated. Run the following command to verify that no other
changes are pending:
terraform plan
Note that the -target option is not suitable for routine use, and is provided
only for exceptional situations such as recovering from errors or mistakes,
or when Terraform specifically suggests to use it as part of an error
message.
run "second"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 2 passed, 0 failed.
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
actual := output.All()
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// There should not be warnings in clean-up
func TestTest_InvalidWarningsInCleanup(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "invalid-cleanup-warnings")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "test"... pass
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Warning: Value for undeclared variable
on main.tftest.hcl line 6, in run "test":
6: validation = "Hello, world!"
The module under test does not declare a variable named "validation", but it
is declared in run block "test".
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 1 passed, 0 failed.
`
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
actual := output.All()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_BadReferences(t *testing.T) {
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "bad-references")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code == 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expectedOut := `main.tftest.hcl... in progress
run "setup"... pass
run "test"... fail
run "finalise"... skip
main.tftest.hcl... tearing down
main.tftest.hcl... fail
providers.tftest.hcl... in progress
run "test"... skip
providers.tftest.hcl... tearing down
providers.tftest.hcl... fail
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Failure! 1 passed, 1 failed, 2 skipped.
`
actualOut := output.Stdout()
if diff := cmp.Diff(actualOut, expectedOut); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expectedErr := `
Error: Reference to unavailable variable
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
on main.tftest.hcl line 15, in run "test":
15: input_one = var.notreal
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
The input variable "notreal" does not exist within this test file.
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
Error: Reference to unknown run block
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
on main.tftest.hcl line 16, in run "test":
16: input_two = run.madeup.response
The run block "madeup" does not exist within this test file.
Error: Reference to unavailable variable
on providers.tftest.hcl line 3, in provider "test":
3: resource_prefix = var.default
The input variable "default" does not exist within this test file.
`
actualErr := output.Stderr()
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_UndefinedVariables(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "variables_undefined_in_config")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code == 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expectedOut := `main.tftest.hcl... in progress
run "test"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`
actualOut := output.Stdout()
if diff := cmp.Diff(actualOut, expectedOut); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
expectedErr := `
Error: Reference to undeclared input variable
on main.tf line 2, in resource "test_resource" "foo":
2: value = var.input
An input variable with the name "input" has not been declared. This variable
can be declared with a variable "input" {} block.
`
actualErr := output.Stderr()
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_VariablesInProviders(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "provider_vars")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "test"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 1 passed, 0 failed.
`
actual := output.All()
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_ExpectedFailuresDuringPlanning(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "expected_failures_during_planning")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code == 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expectedOut := `check.tftest.hcl... in progress
run "check_passes"... pass
check.tftest.hcl... tearing down
check.tftest.hcl... pass
input.tftest.hcl... in progress
run "input_failure"... fail
Warning: Expected failure while planning
A custom condition within var.input failed during the planning stage and
prevented the requested apply operation. While this was an expected failure,
the apply operation could not be executed and so the overall test case will
be marked as a failure and the original diagnostic included in the test
report.
run "no_run"... skip
input.tftest.hcl... tearing down
input.tftest.hcl... fail
output.tftest.hcl... in progress
run "output_failure"... fail
Warning: Expected failure while planning
on output.tftest.hcl line 13, in run "output_failure":
13: output.output,
A custom condition within output.output failed during the planning stage and
prevented the requested apply operation. While this was an expected failure,
the apply operation could not be executed and so the overall test case will
be marked as a failure and the original diagnostic included in the test
report.
output.tftest.hcl... tearing down
output.tftest.hcl... fail
resource.tftest.hcl... in progress
run "resource_failure"... fail
Warning: Expected failure while planning
A custom condition within test_resource.resource failed during the planning
stage and prevented the requested apply operation. While this was an expected
failure, the apply operation could not be executed and so the overall test
case will be marked as a failure and the original diagnostic included in the
test report.
resource.tftest.hcl... tearing down
resource.tftest.hcl... fail
Failure! 1 passed, 3 failed, 1 skipped.
`
actualOut := output.Stdout()
terraform: Move variable validation into its separate node type The parent commit added a new node to the graph for each existing input variable node, but that node didn't do anything yet. Now those new nodes are actually responsible for checking the variable's validation rules, instead of that being included in the same graph node that decides the variable's final value. This also refines the validation node graph transformer to only insert a validation node if there's at least one rule to check, which should avoid bloating the graph for anyone who isn't actually using input variable validation. Therefore the plan graph builder tests revert back to their old expected result because that input configuration does not include any validation rules. For the moment this just moves around functionality that already existed and doesn't add anything new, but having the validation handled in a separate graph node means that we can evaluate the validation rules in the EvalContext of the module where the variable was declared instead of where it was defined, and so a future commit will allow -- initially just as an experiment -- validation rules that can refer to other objects in the module where the variable is declared, including the values of other input variables. As a nice bonus, this fixed a small bug in "terraform test" where problems with variable values defined in the test run were reported as problems with the variable declaration, rather than with the expression assigning the value to it. Since it's now NodeRootVariable that decides the definition source range, rather than the validation-rule-checking helper, it has access to the source range provided in a terraform.InputValue object, which was already populated correctly for per-test-run definitions.
2024-04-04 19:35:23 -04:00
if diff := cmp.Diff(expectedOut, actualOut); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
expectedErr := `
Error: Invalid value for variable
terraform: Move variable validation into its separate node type The parent commit added a new node to the graph for each existing input variable node, but that node didn't do anything yet. Now those new nodes are actually responsible for checking the variable's validation rules, instead of that being included in the same graph node that decides the variable's final value. This also refines the validation node graph transformer to only insert a validation node if there's at least one rule to check, which should avoid bloating the graph for anyone who isn't actually using input variable validation. Therefore the plan graph builder tests revert back to their old expected result because that input configuration does not include any validation rules. For the moment this just moves around functionality that already existed and doesn't add anything new, but having the validation handled in a separate graph node means that we can evaluate the validation rules in the EvalContext of the module where the variable was declared instead of where it was defined, and so a future commit will allow -- initially just as an experiment -- validation rules that can refer to other objects in the module where the variable is declared, including the values of other input variables. As a nice bonus, this fixed a small bug in "terraform test" where problems with variable values defined in the test run were reported as problems with the variable declaration, rather than with the expression assigning the value to it. Since it's now NodeRootVariable that decides the definition source range, rather than the validation-rule-checking helper, it has access to the source range provided in a terraform.InputValue object, which was already populated correctly for per-test-run definitions.
2024-04-04 19:35:23 -04:00
on input.tftest.hcl line 5, in run "input_failure":
5: input = "bcd"
var.input is "bcd"
input must contain the character 'a'
This was checked by the validation rule at main.tf:5,3-13.
Error: Module output value precondition failed
on main.tf line 33, in output "output":
33: condition = strcontains(test_resource.resource.value, "d")
test_resource.resource.value is "abc"
input must contain the character 'd'
Error: Resource postcondition failed
on main.tf line 16, in resource "test_resource" "resource":
16: condition = strcontains(self.value, "b")
self.value is "acd"
input must contain the character 'b'
`
actualErr := output.Stderr()
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_MissingExpectedFailuresDuringApply(t *testing.T) {
// Test asserting that the test run fails, but not errors out, when expected failures are not present during apply.
// This lets subsequent runs continue to execute and the file to be marked as failed.
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "expect_failures_during_apply")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code == 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expectedOut := `main.tftest.hcl... in progress
run "test"... fail
run "follow-up"... pass
Warning: Value for undeclared variable
on main.tftest.hcl line 16, in run "follow-up":
16: input = "does not matter"
The module under test does not declare a variable named "input", but it is
declared in run block "follow-up".
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 1 passed, 1 failed.
`
actualOut := output.Stdout()
if diff := cmp.Diff(expectedOut, actualOut); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
expectedErr := `
Error: Missing expected failure
on main.tftest.hcl line 7, in run "test":
7: output.output
The checkable object, output.output, was expected to report an error but did
not.
`
actualErr := output.Stderr()
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_UnknownAndNulls(t *testing.T) {
tcs := map[string]struct {
code int
stdout string
stderr string
}{
"null_value_in_assert": {
code: 1,
stdout: `main.tftest.hcl... in progress
run "first"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`,
stderr: `
Error: Test assertion failed
on main.tftest.hcl line 8, in run "first":
8: condition = test_resource.resource.value == output.null_output
Diff:
--- actual
+++ expected
- "bar"
+ null
this is always going to fail
`,
},
"null_value_in_vars": {
code: 1,
stdout: `fail.tftest.hcl... in progress
run "first"... pass
run "second"... fail
fail.tftest.hcl... tearing down
fail.tftest.hcl... fail
pass.tftest.hcl... in progress
run "first"... pass
run "second"... pass
pass.tftest.hcl... tearing down
pass.tftest.hcl... pass
Failure! 3 passed, 1 failed.
`,
stderr: `
Error: Required variable not set
on fail.tftest.hcl line 11, in run "second":
11: interesting_input = run.first.null_output
The given value is not suitable for var.interesting_input defined at
main.tf:7,1-29: required variable may not be set to null.
`,
},
"unknown_value_in_assert": {
code: 1,
stdout: `main.tftest.hcl... in progress
run "one"... pass
run "two"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 1 passed, 1 failed.
`,
stderr: fmt.Sprintf(`
Error: Unknown condition value
on main.tftest.hcl line 8, in run "two":
8: condition = output.destroy_fail == run.one.destroy_fail
output.destroy_fail is false
run.one.destroy_fail is a bool
Condition expression could not be evaluated at this time. This means you have
executed a %s block with %s and one of the values your
condition depended on is not known until after the plan has been applied.
Either remove this value from your condition, or execute an %s command
from this %s block. Alternatively, if there is an override for this value,
you can make it available during the plan phase by setting %s in the %s block.
`, "`run`", "`command = plan`", "`apply`", "`run`", "`override_during =\nplan`", "`override_`"),
},
"unknown_value_in_vars": {
code: 1,
stdout: `main.tftest.hcl... in progress
run "one"... pass
run "two"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 1 passed, 1 failed.
`,
stderr: `
Error: Reference to unknown value
on main.tftest.hcl line 8, in run "two":
8: destroy_fail = run.one.destroy_fail
The value for run.one.destroy_fail is unknown. Run block "one" is executing a
"plan" operation, and the specified output value is only known after apply.
`,
},
"nested_unknown_values": {
code: 1,
stdout: `main.tftest.hcl... in progress
run "first"... pass
run "second"... pass
run "third"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 2 passed, 1 failed.
`,
stderr: `
Error: Reference to unknown value
on main.tftest.hcl line 31, in run "third":
31: input = run.second
The value for run.second is unknown. Run block "second" is executing a "plan"
operation, and the specified output value is only known after apply.
`,
},
}
for name, tc := range tcs {
t.Run(name, func(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", name)), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != tc.code {
t.Errorf("expected return code %d but got %d", tc.code, code)
}
expectedOut := tc.stdout
actualOut := output.Stdout()
if diff := cmp.Diff(expectedOut, actualOut); len(diff) > 0 {
t.Errorf("unexpected output\n\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
expectedErr := tc.stderr
actualErr := output.Stderr()
if diff := cmp.Diff(expectedErr, actualErr); len(diff) > 0 {
t.Errorf("unexpected output\n\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
})
}
}
func TestTest_SensitiveInputValues(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "sensitive_input_values")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color", "-verbose"})
output := done(t)
if code != 1 {
t.Errorf("expected status code 1 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "setup"... pass
Outputs:
password = (sensitive value)
run "test"... pass
# test_resource.resource:
resource "test_resource" "resource" {
destroy_fail = false
id = "9ddca5a9"
value = (sensitive value)
write_only = (write-only attribute)
}
Outputs:
password = (sensitive value)
run "test_failed"... fail
# test_resource.resource:
resource "test_resource" "resource" {
destroy_fail = false
id = "9ddca5a9"
value = (sensitive value)
write_only = (write-only attribute)
}
Outputs:
password = (sensitive value)
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 2 passed, 1 failed.
`
expectedErr := `
Error: Test assertion failed
on main.tftest.hcl line 27, in run "test_failed":
27: condition = var.complex == {
28: foo = "bar"
29: baz = test_resource.resource.id
30: }
LHS:
{
"baz": "(sensitive value)",
"foo": "bar"
}
RHS:
{
"baz": "9ddca5a9",
"foo": "bar"
}
Diff:
--- actual
+++ expected
{
- "baz": "(sensitive value)",
+ "baz": "9ddca5a9",
"foo": "bar"
}
test_resource.resource.id is "9ddca5a9"
var.complex is {
"baz": "(sensitive value)",
"foo": "bar"
}
expected to fail
`
actual := output.Stdout()
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if diff := cmp.Diff(output.Stderr(), expectedErr); len(diff) > 0 {
t.Errorf("stderr didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, output.Stderr(), diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
// This test takes around 10 seconds to complete, as we're testing the progress
// updates that are printed every 2 seconds. Sorry!
func TestTest_LongRunningTest(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "long_running")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
actual := output.All()
expected := `main.tftest.hcl... in progress
run "test"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 1 passed, 0 failed.
`
if code != 0 {
t.Errorf("expected return code %d but got %d", 0, code)
}
if diff := cmp.Diff(expected, actual); len(diff) > 0 {
t.Errorf("unexpected output\n\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
}
// This test takes around 10 seconds to complete, as we're testing the progress
// updates that are printed every 2 seconds. Sorry!
func TestTest_LongRunningTestJSON(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "long_running")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
code := c.Run([]string{"-json"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
actual := output.All()
var messages []string
for ix, line := range strings.Split(actual, "\n") {
if len(line) == 0 {
// Skip empty lines.
continue
}
if ix == 0 {
// skip the first one, it's version information
continue
}
var obj map[string]interface{}
if err := json.Unmarshal([]byte(line), &obj); err != nil {
t.Errorf("failed to unmarshal returned line: %s", line)
continue
}
// Remove the timestamp as it changes every time.
delete(obj, "@timestamp")
if obj["type"].(string) == "test_run" {
// Then we need to delete the `elapsed` field from within the run
// as it'll cause flaky tests.
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
run := obj["test_run"].(map[string]interface{})
if run["progress"].(string) != "complete" {
delete(run, "elapsed")
}
}
message, err := json.Marshal(obj)
if err != nil {
t.Errorf("failed to remarshal returned line: %s", line)
continue
}
messages = append(messages, string(message))
}
expected := []string{
`{"@level":"info","@message":"Found 1 file and 1 run block","@module":"terraform.ui","test_abstract":{"main.tftest.hcl":["test"]},"type":"test_abstract"}`,
`{"@level":"info","@message":"main.tftest.hcl... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","test_file":{"path":"main.tftest.hcl","progress":"starting"},"type":"test_file"}`,
`{"@level":"info","@message":" \"test\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"starting","run":"test"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"running","run":"test"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test\"... in progress","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"running","run":"test"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test\"... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"complete","run":"test","status":"pass"},"type":"test_run"}`,
`{"@level":"info","@message":"main.tftest.hcl... tearing down","@module":"terraform.ui","@testfile":"main.tftest.hcl","test_file":{"path":"main.tftest.hcl","progress":"teardown"},"type":"test_file"}`,
`{"@level":"info","@message":" \"test\"... tearing down","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"teardown","run":"test"},"type":"test_run"}`,
`{"@level":"info","@message":" \"test\"... tearing down","@module":"terraform.ui","@testfile":"main.tftest.hcl","@testrun":"test","test_run":{"path":"main.tftest.hcl","progress":"teardown","run":"test"},"type":"test_run"}`,
`{"@level":"info","@message":"main.tftest.hcl... pass","@module":"terraform.ui","@testfile":"main.tftest.hcl","test_file":{"path":"main.tftest.hcl","progress":"complete","status":"pass"},"type":"test_file"}`,
`{"@level":"info","@message":"Success! 1 passed, 0 failed.","@module":"terraform.ui","test_summary":{"errored":0,"failed":0,"passed":1,"skipped":0,"status":"pass"},"type":"test_summary"}`,
}
if code != 0 {
t.Errorf("expected return code %d but got %d", 0, code)
}
if diff := cmp.Diff(expected, messages); len(diff) > 0 {
t.Errorf("unexpected output\n\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", strings.Join(expected, "\n"), strings.Join(messages, "\n"), diff)
}
}
func TestTest_InvalidOverrides(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "invalid-overrides")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "setup"... pass
Warning: Invalid override target
on main.tftest.hcl line 39, in run "setup":
39: target = test_resource.absent_five
The override target test_resource.absent_five does not exist within the
configuration under test. This could indicate a typo in the target address or
an unnecessary override.
run "test"... pass
Warning: Invalid override target
on main.tftest.hcl line 45, in run "test":
45: target = module.setup.test_resource.absent_six
The override target module.setup.test_resource.absent_six does not exist
within the configuration under test. This could indicate a typo in the target
address or an unnecessary override.
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Warning: Invalid override target
on main.tftest.hcl line 4, in mock_provider "test":
4: target = test_resource.absent_one
The override target test_resource.absent_one does not exist within the
configuration under test. This could indicate a typo in the target address or
an unnecessary override.
(and 3 more similar warnings elsewhere)
Success! 2 passed, 0 failed.
`
actual := output.All()
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_InvalidConfig(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "invalid_config")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output := done(t)
if code != 1 {
t.Errorf("expected status code ! but got %d", code)
}
expectedOut := `main.tftest.hcl... in progress
run "test"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`
expectedErr := `
Error: Failed to load plugin schemas
Error while loading schemas for plugin components: Failed to obtain provider
schema: Could not load the schema for provider
registry.terraform.io/hashicorp/test: failed to instantiate provider
"registry.terraform.io/hashicorp/test" to obtain schema: fork/exec
.terraform/providers/registry.terraform.io/hashicorp/test/1.0.0/%s/terraform-provider-test_1.0.0:
permission denied..
`
expectedErr = fmt.Sprintf(expectedErr, runtime.GOOS+"_"+runtime.GOARCH)
out := output.Stdout()
err := output.Stderr()
if diff := cmp.Diff(out, expectedOut); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, out, diff)
}
if diff := cmp.Diff(err, expectedErr); len(diff) > 0 {
t.Errorf("error didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, err, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
// TestTest_ValidateBackendConfiguration tests validation of how backends are declared in test files:
// * it's not valid to re-use the same backend config (i.e the same state file)
// * it's not valid to use a deprecated backend type
// * it's not valid to use a non-existent backend type
//
// Backend validation performed in the command package is dependent on the internal/backend/init package,
// which cannot be imported in configuration parsing packages without creating an import cycle.
func TestTest_ReusedBackendConfiguration(t *testing.T) {
testCases := map[string]struct {
dirName string
expectErr string
}{
"validation detects when backend config is reused by runs using different user-supplied state key value": {
dirName: "reused-backend-config",
expectErr: `
Error: Repeat use of the same backend block
on main.tftest.hcl line 12, in run "test_2":
12: backend "local" {
The run "test_2" contains a backend configuration that's already been used in
run "test_1". Sharing the same backend configuration between separate runs
will result in conflicting state updates.
`,
},
"validation detects when backend config is reused by runs using different implicit state key (corresponding to root and a child module) ": {
dirName: "reused-backend-config-child-modules",
expectErr: `
Error: Repeat use of the same backend block
on main.tftest.hcl line 19, in run "test_2":
19: backend "local" {
The run "test_2" contains a backend configuration that's already been used in
run "test_1". Sharing the same backend configuration between separate runs
will result in conflicting state updates.
`,
},
"validation detects when a deprecated backend type is used": {
dirName: "removed-backend-type",
expectErr: `
Error: Unsupported backend type
on main.tftest.hcl line 7, in run "test_removed_backend":
7: backend "etcd" {
The "etcd" backend is not supported in Terraform v1.3 or later.
`,
},
"validation detects when a non-existent backend type": {
dirName: "non-existent-backend-type",
expectErr: `
Error: Unsupported backend type
on main.tftest.hcl line 7, in run "test_invalid_backend":
7: backend "foobar" {
There is no backend type named "foobar".
`,
},
}
for tn, tc := range testCases {
t.Run(tn, func(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", tc.dirName)), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
init := &InitCommand{
Meta: meta,
}
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output = done(t)
// Assertions
if code != 1 {
t.Errorf("expected status code 1 but got %d", code)
}
if diff := cmp.Diff(output.All(), tc.expectErr); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", tc.expectErr, output.All(), diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
})
}
}
// When there is no starting state, state is created by the run containing the backend block
func TestTest_UseOfBackends_stateCreatedByBackend(t *testing.T) {
dirName := "valid-use-local-backend/no-prior-state"
resourceId := "12345"
expectedState := `test_resource.foobar:
ID = 12345
provider = provider["registry.terraform.io/hashicorp/test"]
destroy_fail = false
value = value-from-run-that-controls-backend
Outputs:
supplied_input_value = value-from-run-that-controls-backend
test_resource_id = 12345`
// SETUP
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", dirName)), td)
t.Chdir(td)
localStatePath := filepath.Join(td, DefaultStateFilename)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
// INIT
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// TEST
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
code := c.Run([]string{"-no-color"})
output := done(t)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// ASSERTIONS
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
stdErr := output.Stderr()
if len(stdErr) > 0 {
t.Fatalf("unexpected error output:\n%s", stdErr)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// State is stored according to the backend block
actualState := testStateRead(t, localStatePath)
if diff := cmp.Diff(actualState.String(), expectedState); len(diff) > 0 {
t.Fatalf("state didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedState, actualState, diff)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if provider.ResourceCount() != 1 {
t.Fatalf("should have deleted all resources on completion except test_resource.a. Instead state contained %b resources: %v", provider.ResourceCount(), provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
val := provider.Store.Get(resourceId)
if val.GetAttr("id").AsString() != resourceId {
t.Errorf("expected resource to have value %q but got %s", resourceId, val.GetAttr("id").AsString())
}
if val.GetAttr("value").AsString() != "value-from-run-that-controls-backend" {
t.Errorf("expected resource to have value 'value-from-run-that-controls-backend' but got %s", val.GetAttr("value").AsString())
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// When there is pre-existing state, the state is used by the run containing the backend block
func TestTest_UseOfBackends_priorStateUsedByBackend(t *testing.T) {
dirName := "valid-use-local-backend/with-prior-state"
resourceId := "53d69028-477d-7ba0-83c3-ff3807e3756f" // This value needs to match the state file in the test fixtures
expectedState := fmt.Sprintf(`test_resource.foobar:
ID = %s
provider = provider["registry.terraform.io/hashicorp/test"]
destroy_fail = false
value = value-from-run-that-controls-backend
Outputs:
supplied_input_value = value-from-run-that-controls-backend
test_resource_id = %s`, resourceId, resourceId)
// SETUP
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", dirName)), td)
t.Chdir(td)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
localStatePath := filepath.Join(td, DefaultStateFilename)
// resource store is like the remote object - needs to be assembled
// to resemble what's in state / the remote object being reused by the test.
resourceStore := &testing_command.ResourceStore{
Data: map[string]cty.Value{
resourceId: cty.ObjectVal(map[string]cty.Value{
"id": cty.StringVal(resourceId),
"interrupt_count": cty.NullVal(cty.Number),
"value": cty.StringVal("value-from-run-that-controls-backend"),
"write_only": cty.NullVal(cty.String),
"create_wait_seconds": cty.NullVal(cty.Number),
"destroy_fail": cty.False,
"destroy_wait_seconds": cty.NullVal(cty.Number),
"defer": cty.NullVal(cty.Bool),
})},
}
provider := testing_command.NewProvider(resourceStore)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// INIT
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// TEST
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
output := done(t)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// ASSERTIONS
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
stdErr := output.Stderr()
if len(stdErr) > 0 {
t.Fatalf("unexpected error output:\n%s", stdErr)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// State is stored according to the backend block
actualState := testStateRead(t, localStatePath)
if diff := cmp.Diff(actualState.String(), expectedState); len(diff) > 0 {
t.Fatalf("state didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedState, actualState, diff)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if provider.ResourceCount() != 1 {
t.Fatalf("should have deleted all resources on completion except test_resource.a. Instead state contained %b resources: %v", provider.ResourceCount(), provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
val := provider.Store.Get(resourceId)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// If the ID hasn't changed then we've used the pre-existing state, instead of remaking the resource
if val.GetAttr("id").AsString() != resourceId {
t.Errorf("expected resource to have value %q but got %s", resourceId, val.GetAttr("id").AsString())
}
if val.GetAttr("value").AsString() != "value-from-run-that-controls-backend" {
t.Errorf("expected resource to have value 'value-from-run-that-controls-backend' but got %s", val.GetAttr("value").AsString())
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// Testing whether a state artifact is made for a run block with a backend or not
//
// Artifacts are made when the cleanup operation errors.
func TestTest_UseOfBackends_whenStateArtifactsAreMade(t *testing.T) {
cases := map[string]struct {
forceError bool
expectedCode int
expectStateManifest bool
}{
"no artifact made when there are no cleanup errors when processing a run block with a backend": {
forceError: false,
expectedCode: 0,
expectStateManifest: false,
},
"artifact made when a cleanup error is forced when processing a run block with a backend": {
forceError: true,
expectedCode: 1,
expectStateManifest: true,
},
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
for tn, tc := range cases {
t.Run(tn, func(t *testing.T) {
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// SETUP
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "valid-use-local-backend/no-prior-state")), td)
t.Chdir(td)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
provider := testing_command.NewProvider(nil)
erroringInvocationNum := 3
applyResourceChangeCount := 0
if tc.forceError {
oldFunc := provider.Provider.ApplyResourceChangeFn
newFunc := func(req providers.ApplyResourceChangeRequest) providers.ApplyResourceChangeResponse {
applyResourceChangeCount++
if applyResourceChangeCount < erroringInvocationNum {
return oldFunc(req)
}
// Given the config in the test fixture used, the 5th call to this function is during cleanup
// Return error to force error diagnostics during cleanup
var diags tfdiags.Diagnostics
return providers.ApplyResourceChangeResponse{
Diagnostics: diags.Append(errors.New("error forced by mock provider")),
}
}
provider.Provider.ApplyResourceChangeFn = newFunc
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// INIT
init := &InitCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// TEST
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
c := &TestCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
code := c.Run([]string{"-no-color"})
// ASSERTIONS
if tc.forceError && (applyResourceChangeCount != erroringInvocationNum) {
t.Fatalf(`Test did not force error as expected. This is because a magic number in the test setup is coupled to the config.
The apply resource change function was invoked %d times but we trigger an error on the %dth time`, applyResourceChangeCount, erroringInvocationNum)
}
if code != tc.expectedCode {
output := done(t)
t.Errorf("expected status code %d but got %d: %s", tc.expectedCode, code, output.All())
}
// State is NOT stored in .terraform/test as a state artifact because
// there haven't been any failures or errors in the tests
manifest, err := teststates.LoadManifest(td, true)
if err != nil {
t.Fatal(err)
}
foundIds := []string{}
for _, file := range manifest.Files {
for _, state := range file.States {
foundIds = append(foundIds, state.ID)
}
}
if len(foundIds) > 0 && !tc.expectStateManifest {
t.Fatalf("found %d state files in .terraform/test when none were expected", len(foundIds))
}
if len(foundIds) == 0 && tc.expectStateManifest {
t.Fatalf("found 0 state files in .terraform/test when they were were expected")
}
})
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_UseOfBackends_validatesUseOfSkipCleanup(t *testing.T) {
cases := map[string]struct {
testDir string
expectCode int
expectErr bool
}{
"cannot set skip_cleanup=false alongside a backend block": {
testDir: "backend-with-skip-cleanup/false",
expectCode: 1,
expectErr: true,
},
"can set skip_cleanup=true alongside a backend block": {
testDir: "backend-with-skip-cleanup/true",
expectCode: 0,
expectErr: false,
},
}
for tn, tc := range cases {
t.Run(tn, func(t *testing.T) {
// SETUP
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", tc.testDir)), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
// INIT
init := &InitCommand{
Meta: meta,
}
code := init.Run([]string{"-no-color"})
output := done(t)
// ASSERTIONS
if code != tc.expectCode {
t.Errorf("expected status code %d but got %d", tc.expectCode, code)
}
stdErr := output.Stderr()
if len(stdErr) == 0 && tc.expectErr {
t.Fatal("expected error output but got none")
}
if len(stdErr) != 0 && !tc.expectErr {
t.Fatalf("did not expect error output but got: %s", stdErr)
}
if provider.ResourceCount() > 0 {
t.Fatalf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
})
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func TestTest_UseOfBackends_failureDuringApply(t *testing.T) {
// SETUP
td := t.TempDir()
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testCopyDir(t, testFixturePath(path.Join("test", "valid-use-local-backend/no-prior-state")), td)
t.Chdir(td)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
localStatePath := filepath.Join(td, DefaultStateFilename)
provider := testing_command.NewProvider(nil)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// Force a failure during apply
provider.Provider.ApplyResourceChangeFn = func(req providers.ApplyResourceChangeRequest) providers.ApplyResourceChangeResponse {
resp := providers.ApplyResourceChangeResponse{}
resp.Diagnostics = resp.Diagnostics.Append(fmt.Errorf("forced error"))
return resp
}
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
AllowExperimentalFeatures: true,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// INIT
init := &InitCommand{
Meta: meta,
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output := done(t)
if code := init.Run(nil); code != 0 {
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// TEST
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
output = done(t)
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// ASSERTIONS
if code != 1 {
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
t.Errorf("expected status code 1 but got %d", code)
}
stdErr := output.Stderr()
if len(stdErr) == 0 {
t.Fatal("expected error output but got none")
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// Resource was not provisioned
if provider.ResourceCount() > 0 {
t.Fatalf("no resources should have been provisioned successfully but got %v", provider.ResourceString())
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
// When there is a failure to apply changes to the test_resource, the resulting state saved via the backend
// only includes the output and lacks any information about the test_resource
actualBackendState := testStateRead(t, localStatePath)
expectedBackendState := `<no state>
Outputs:
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
supplied_input_value = value-from-run-that-controls-backend
test_resource_id = 12345`
if diff := cmp.Diff(actualBackendState.String(), expectedBackendState); len(diff) > 0 {
t.Fatalf("state didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedBackendState, actualBackendState, diff)
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
expectedStates := map[string][]string{} // empty
actualStates := statesFromManifest(t, td)
// No state artifacts are made: Verify the states in the manifest
if diff := cmp.Diff(expectedStates, removeOutputs(actualStates)); diff != "" {
t.Fatalf("unexpected states: %s", diff)
}
}
func TestTest_RunBlocksInProviders(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "provider_runs")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
test := &TestCommand{
Meta: meta,
}
code := test.Run([]string{"-no-color"})
output := done(t)
if code != 0 {
t.Errorf("expected status code 0 but got %d", code)
}
expected := `main.tftest.hcl... in progress
run "setup"... pass
run "main"... pass
main.tftest.hcl... tearing down
main.tftest.hcl... pass
Success! 2 passed, 0 failed.
`
actual := output.All()
if diff := cmp.Diff(actual, expected); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expected, actual, diff)
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
}
func TestTest_RunBlocksInProviders_BadReferences(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "provider_runs_invalid")), td)
t.Chdir(td)
store := &testing_command.ResourceStore{
Data: make(map[string]cty.Value),
}
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: &testingOverrides{
Providers: map[addrs.Provider]providers.Factory{
addrs.NewDefaultProvider("test"): func() (providers.Interface, error) {
return testing_command.NewProvider(store).Provider, nil
},
},
},
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code 0 but got %d: %s", code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
test := &TestCommand{
Meta: meta,
}
code := test.Run([]string{"-no-color"})
output := done(t)
if code != 1 {
t.Errorf("expected status code 1 but got %d", code)
}
expectedOut := `missing_run_block.tftest.hcl... in progress
run "main"... skip
missing_run_block.tftest.hcl... tearing down
missing_run_block.tftest.hcl... fail
unused_provider.tftest.hcl... in progress
run "main"... pass
unused_provider.tftest.hcl... tearing down
unused_provider.tftest.hcl... pass
Failure! 1 passed, 0 failed, 1 skipped.
`
actualOut := output.Stdout()
if diff := cmp.Diff(actualOut, expectedOut); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, actualOut, diff)
}
expectedErr := `
Error: Reference to unknown run block
on missing_run_block.tftest.hcl line 2, in provider "test":
2: resource_prefix = run.missing.resource_directory
The run block "missing" does not exist within this test file.
`
actualErr := output.Stderr()
if diff := cmp.Diff(actualErr, expectedErr); len(diff) > 0 {
t.Errorf("output didn't match expected:\nexpected:\n%s\nactual:\n%s\ndiff:\n%s", expectedErr, actualErr, diff)
}
if len(store.Data) > 0 {
t.Errorf("should have deleted all resources on completion but left %v", len(store.Data))
}
}
func TestTest_JUnitOutput(t *testing.T) {
tcs := map[string]struct {
path string
code int
wantFilename string
}{
"can create XML for a single file with 1 pass, 1 fail": {
path: "junit-output/1pass-1fail",
wantFilename: "expected-output.xml",
code: 1, // Test failure
},
"can create XML for multiple files with 1 pass each": {
path: "junit-output/multiple-files",
wantFilename: "expected-output.xml",
code: 0,
},
"can display a test run's errors under the equivalent test case element": {
path: "junit-output/missing-provider",
wantFilename: "expected-output.xml",
code: 1, // Test error
},
}
for tn, tc := range tcs {
t.Run(tn, func(t *testing.T) {
// Setup test
td := t.TempDir()
testPath := path.Join("test", tc.path)
testCopyDir(t, testFixturePath(testPath), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
view, done := testView(t)
c := &TestCommand{
Meta: Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
View: view,
},
}
// Run command with -junit-xml=./output.xml flag
outputFile := fmt.Sprintf("%s/output.xml", td)
code := c.Run([]string{fmt.Sprintf("-junit-xml=%s", outputFile), "-no-color"})
done(t)
// Assertions
if code != tc.code {
t.Errorf("expected status code %d but got %d", tc.code, code)
}
actualOut, err := os.ReadFile(outputFile)
if err != nil {
t.Fatalf("error opening XML file: %s", err)
}
expectedOutputFile := fmt.Sprintf("%s/%s", td, tc.wantFilename)
expectedOutput, err := os.ReadFile(expectedOutputFile)
if err != nil {
t.Fatalf("error opening XML file: %s", err)
}
// actual output will include timestamps and test duration data, which isn't deterministic; redact it for comparison
timeRegexp := regexp.MustCompile(`time="[^"]+"`)
actualOut = timeRegexp.ReplaceAll(actualOut, []byte("time=\"TIME_REDACTED\""))
timestampRegexp := regexp.MustCompile(`timestamp="[^"]+"`)
actualOut = timestampRegexp.ReplaceAll(actualOut, []byte("timestamp=\"TIMESTAMP_REDACTED\""))
if !bytes.Equal(actualOut, expectedOutput) {
t.Fatalf("wanted XML:\n%s\n got XML:\n%s\ndiff:%s\n", string(expectedOutput), string(actualOut), cmp.Diff(expectedOutput, actualOut))
}
if provider.ResourceCount() > 0 {
t.Errorf("should have deleted all resources on completion but left %v", provider.ResourceString())
}
})
}
}
func TestTest_ReferencesIntoIncompletePlan(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "expect-failures-assertions")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 0, code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
if code != 1 {
t.Errorf("expected status code %d but got %d", 0, code)
}
output := done(t)
out, err := output.Stdout(), output.Stderr()
expectedOut := `main.tftest.hcl... in progress
run "fail"... fail
main.tftest.hcl... tearing down
main.tftest.hcl... fail
Failure! 0 passed, 1 failed.
`
if diff := cmp.Diff(out, expectedOut); len(diff) > 0 {
t.Errorf("expected:\n%s\nactual:\n%s\ndiff:\n%s", expectedOut, out, diff)
}
if !strings.Contains(err, "Reference to uninitialized resource") {
t.Errorf("missing reference to uninitialized resource error: \n%s", err)
}
if !strings.Contains(err, "Reference to uninitialized local") {
t.Errorf("missing reference to uninitialized local error: \n%s", err)
}
}
func TestTest_ReferencesIntoTargetedPlan(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "invalid-reference-with-target")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 0, code, output.All())
}
// Reset the streams for the next command.
streams, done = terminal.StreamsForTesting(t)
meta.Streams = streams
meta.View = views.NewView(streams)
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color"})
if code != 1 {
t.Errorf("expected status code %d but got %d", 0, code)
}
output := done(t)
err := output.Stderr()
if !strings.Contains(err, "Reference to uninitialized variable") {
t.Errorf("missing reference to uninitialized variable error: \n%s", err)
}
}
// https://github.com/hashicorp/terraform/issues/37546
func TestTest_TeardownOrder(t *testing.T) {
td := t.TempDir()
testCopyDir(t, testFixturePath(path.Join("test", "rds_shared_subnet")), td)
t.Chdir(td)
provider := testing_command.NewProvider(nil)
providerSource, close := newMockProviderSource(t, map[string][]string{
"test": {"1.0.0"},
})
defer close()
streams, done := terminal.StreamsForTesting(t)
view := views.NewView(streams)
ui := new(cli.MockUi)
meta := Meta{
testingOverrides: metaOverridesForProvider(provider.Provider),
Ui: ui,
View: view,
Streams: streams,
ProviderSource: providerSource,
}
init := &InitCommand{
Meta: meta,
}
if code := init.Run(nil); code != 0 {
output := done(t)
t.Fatalf("expected status code %d but got %d: %s", 0, code, output.All())
}
c := &TestCommand{
Meta: meta,
}
code := c.Run([]string{"-no-color", "-json"})
if code != 0 {
t.Errorf("expected status code %d but got %d", 0, code)
}
output := done(t).All()
// Parse the JSON output to check teardown order
var setupTeardownStart time.Time
var lastRunTeardownStart time.Time
for line := range strings.SplitSeq(output, "\n") {
if strings.Contains(line, `"progress":"teardown"`) {
var obj map[string]any
if err := json.Unmarshal([]byte(line), &obj); err != nil {
t.Fatal(err)
}
if strings.Contains(line, `"setup_tests"`) {
if ts, ok := obj["@timestamp"].(string); ok {
// record the first time that the setup teardown appears in the output
if setupTeardownStart.IsZero() {
parsedTime, _ := time.Parse(time.RFC3339, ts)
setupTeardownStart = parsedTime
}
}
} else {
if ts, ok := obj["@timestamp"].(string); ok {
parsedTime, _ := time.Parse(time.RFC3339, ts)
// record the last time that a run's teardown appears in the output
if parsedTime.After(lastRunTeardownStart) {
lastRunTeardownStart = parsedTime
}
}
}
}
}
// all runs should have been down with teardown before the setup starts
if lastRunTeardownStart.After(setupTeardownStart) {
t.Fatalf("setup is tearing down before dependants are done: \n %s", output)
}
if provider.ResourceCount() > 0 {
t.Logf("Resources remaining after test completion (this might indicate the teardown issue): %v", provider.ResourceString())
}
}
// testModuleInline takes a map of path -> config strings and yields a config
// structure with those files loaded from disk
func testModuleInline(t *testing.T, sources map[string]string) (*configs.Config, string, func()) {
t.Helper()
cfgPath := t.TempDir()
for path, configStr := range sources {
dir := filepath.Dir(path)
if dir != "." {
err := os.MkdirAll(filepath.Join(cfgPath, dir), os.FileMode(0777))
if err != nil {
t.Fatalf("Error creating subdir: %s", err)
}
}
// Write the configuration
cfgF, err := os.Create(filepath.Join(cfgPath, path))
if err != nil {
t.Fatalf("Error creating temporary file for config: %s", err)
}
_, err = io.Copy(cfgF, strings.NewReader(configStr))
cfgF.Close()
if err != nil {
t.Fatalf("Error creating temporary file for config: %s", err)
}
}
loader, cleanup := configload.NewLoaderForTests(t)
// Test modules usually do not refer to remote sources, and for local
// sources only this ultimately just records all of the module paths
// in a JSON file so that we can load them below.
inst := initwd.NewModuleInstaller(loader.ModulesDir(), loader, registry.NewClient(nil, nil), nil)
_, instDiags := inst.InstallModules(context.Background(), cfgPath, "tests", true, false, initwd.ModuleInstallHooksImpl{})
if instDiags.HasErrors() {
t.Fatal(instDiags.Err())
}
// Since module installer has modified the module manifest on disk, we need
// to refresh the cache of it in the loader.
if err := loader.RefreshModules(); err != nil {
t.Fatalf("failed to refresh modules after installation: %s", err)
}
rootMod, hclDiags := loader.LoadRootModuleWithTests(cfgPath, "tests")
if hclDiags.HasErrors() {
t.Fatal(hclDiags.Error())
}
config, diags := terraform.BuildConfigWithGraph(
rootMod,
loader.ModuleWalker(),
nil,
configs.MockDataLoaderFunc(loader.LoadExternalMockData),
)
if diags.HasErrors() {
t.Fatal(diags.Err())
}
return config, cfgPath, func() {
cleanup()
}
}
Implement controlling destroy functionality within Terraform Test (#37359) * Add ability to parse backend blocks present in a test file's run blocks, validate configuration (#36541) * Add ability to parse backend blocks from a run block * Add validation to avoid multiple backend blocks across run blocks that use the same internal state file. Update tests. * Add validation to avoid multiple backend blocks within a single run block. Update tests. * Remove use of quotes in diagnostic messages * Add validation to avoid backend blocks being used in plan run blocks. Update tests. * Correct local backend blocks in new test fixtures * Add test to show that different test files can use same backend block for same state key. * Add validation to enforce state-storage backend types are used * Remove TODO comment We only need to consider one file at a time when checking if a state_key already has a backend associated with it; parallelism in `terraform test` is scoped down to individual files. * Add validation to assert that the backend block must be in the first apply command for an internal state * Consolidate backend block validation inside a single if statement * Add initial version of validation that ensures a backend isn't re-used within a file * Explicitly set the state_key at the point of parsing the config TODO: What should be done with method (moduletest.Run).GetStateKey? * Update test fixture now that reusing backend configs has been made invalid * Add automated test showing validation of reused configuration blocks * Skip test due to flakiness, minor change to test config naming * Update test so it tolerates non-deterministic order run blocks are evaluated in * Remove unnecessary value assignment to r.StateKey * Replace use of GetStateKey() with accessing the state key that's now set during test config parsing * Fix bug so that run blocks using child modules get the correct state key set at parsing time * Update acceptance test to also cover scenario where root and child module state keys are in use * Update test name * Add newline to regex * Ensure consistent place where repeat backend error is raised from * Write leftover test state(s) to file (#36614) * Add additional validation that the backend used in a run is a supported type (#36648) * Prevent test run when leftover state data is present (#36685) * `test`: Set the initial state for a state files from a backend, allow the run that defines a backend to write state to the backend (#36646) * Allow use of backend block to set initial state for a state key * Note about alternative place to keep 'backend factories' * Allow the run block defining the backend to write state to it * Fix rebase * Change to accessing backend init functions via ContextOpts * Add tests demonstrating how runs containing backend blocks use and update persisted state * Fix test fixture * Address test failure due to trouble opening the state file This problem doesn't happen on MacOS, so I assume is due to the Linux environment of GitHub runners. * Fix issue with paths properly I hope * Fix defect in test assertion * Pivot back to approach introduced in 4afc3d7 * Let failing tests write to persistent state, add test case covering that. I split the acceptance tests into happy/unhappy paths for this, which required some of the helper functions' declarations to be raised up to package-level. * Change how we update internal state files, so that information about the associated backend is never lost * Fix UpdateStateFile * Ensure that the states map set by TestStateTransformer associates a backend with the correct run. * Misc spelling fixes in comments and a log * Replace state get/set functions with existing helpers (#36747) * Replace state get/set functions with existing helpers * Compare to string representation of state * Compare to string representation of state * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * Add validation to enforce skip_cleanup=false cannot be used with backend blocks (#36857) * Integrate use of backend blocks in tests with skip_cleanup feature (#36848) * Fix nil pointer error, update test to not be table-driven * Make using a backend block implicitly set skip_cleanup to true * Stop state artefacts being created when a backend is in use and no cleanup errors have occurred * Return diagnostics so calling code knows if cleanup experienced issues or not * Update tests to show that when cleanup fails a state artefact is created * Add comment about why diag not returned * Bug fix - actually pull in the state from the state manager! * Split and simplify (?) tests to show the backend block can create and/or reuse prior state * Update test to use new fixtures, assert about state artefact. Fix nil pointer * Update test fixture in use, add guardrail for flakiness of forced error during cleanup * Refactor so resource ID set in only one place * Add documentation for using a `backend` block during `test` (#36832) * Add backend as a documented block in a run block * Add documentation about backend blocks in run blocks. * Make the relationship between backends and state keys more clear, other improvements * More test documentation (#36838) * Terraform Test: cleanup command (#36847) * Allow cleanup of states that depend on prior runs outputs (#36902) * terraform test: refactor graph edge calculation * create fake run block nodes during cleanup operation * tidy up TODOs * fix tests * remove old changes * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * Improve diagnostics around skip_cleanup conflicts (#37385) * Improve diagnostics around skip_cleanup conflicts * remove unused dynamic node * terraform test: refactor manifest file for simplicity (#37412) * test: refactor apply and plan functions so no run block is needed * terraform test: write and load state manifest files * Terraform Test: Allow skipping cleanup of entire test file or individual run blocks (#36729) * terraform test: add support for skip_cleanup attr * terraform test: add cleanup command * terraform test: add backend blocks * pause * fix tests * remove commented code * terraform test: make controlling destroy functionality experimental (#37419) * address comments * Update internal/moduletest/graph/node_state_cleanup.go Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> --------- Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> * add experimental changelog entries --------- Co-authored-by: Sarah French <15078782+SarahFrench@users.noreply.github.com> Co-authored-by: Samsondeen <40821565+dsa0x@users.noreply.github.com> Co-authored-by: Samsondeen Dare <samsondeen.dare@hashicorp.com>
2025-09-10 11:22:20 -04:00
func statesFromManifest(t *testing.T, td string) map[string][]string {
manifest, err := teststates.LoadManifest(td, true)
if err != nil {
t.Fatal(err)
}
states := make(map[string][]string)
// Verify the states in the manifest
for fileName, file := range manifest.Files {
for name, state := range file.States {
sm := statemgr.NewFilesystem(manifest.StateFilePath(state.ID))
if err := sm.RefreshState(); err != nil {
t.Fatalf("error when reading state file: %s", err)
}
state := sm.State()
// If the state is nil, then the test cleaned up the state
if state == nil || state.Empty() {
continue
}
var resources []string
for _, module := range state.Modules {
for _, resource := range module.Resources {
resources = append(resources, resource.Addr.String())
}
}
for _, output := range state.RootOutputValues {
resources = append(resources, output.Addr.String())
}
if len(resources) == 0 {
continue
}
sort.Strings(resources)
states[strings.TrimSuffix(fileName, ".tftest.hcl")+"."+name] = resources
}
}
return states
}
func equalIgnoreOrder() cmp.Option {
less := func(a, b string) bool { return a < b }
return cmpopts.SortSlices(less)
}
func removeOutputs(states map[string][]string) map[string][]string {
for k, v := range states {
new_v := make([]string, 0, len(v))
for _, s := range v {
if !strings.HasPrefix(s, "output.") {
new_v = append(new_v, s)
}
}
if len(new_v) == 0 {
delete(states, k)
continue
}
states[k] = new_v
}
return states
}