* Update mmctl-test-template.yml to conditionally execute docker login for mattermost repository owner
* Update build-server-image.yml and server-test-template.yml to conditionally execute docker login for mattermost repository owner
* Update workflow files to conditionally execute docker login for pull requests from mattermost/mattermost
* Enhance CI workflows to conditionally execute jobs for pull requests from mattermost/mattermost
* Remove conditional check for docker login in CI workflows to streamline execution across all pull requests
* Add conditional check for docker login in CI workflows for pull requests from mattermost/mattermost
* Refactor CI workflows to gracefully skip tests for external pull requests, enhancing clarity and execution flow
* revert checks
* skip fips testing for forks
* Fix fork detection in FIPS tests to allow push events
The previous condition `github.repository_owner == 'mattermost'` was
incorrect as it would always be false for forks. Updated to properly
detect forks using `github.event.pull_request.head.repo.full_name == github.repository`
and allow FIPS tests to run on push events.
---------
Co-authored-by: Jesse Hallam <jesse@mattermost.com>
* pin to ubuntu-24.04
* always use FIPS compatible Postgres settings
* use sha256 for remote cluster IDs
* use sha256 for client config hash
* rework S3 backend to be FIPS compatible
* skip setup-node during build, since already in container
* support FIPS builds
* Dockerfile for FIPS image, using glibc-openssl-fips
* workaround entrypoint inconsistencies
* authenticate to DockerHub
* fix FIPS_ENABLED, add test-mmctl-fips
* decouple check-mattermost-vet from test/build steps
* fixup! decouple check-mattermost-vet from test/build steps
* only build-linux-amd64 for fips
* rm entrypoint workaround
* tweak comment grammar
* rm unused Dockerfile.fips (for now)
* ignore gpg import errors, since would fail later anyway
* for fips, only make package-linux-amd64
* set FIPS_ENABLED for build step
* Add a FIPS-specific list of prepackaged plugins
Note that the names are still temporary, since they are not uploaded to
S3 yet. We may need to tweak them when that happens.
* s/golangci-lint/check-style/
This ensures we run all the `check-style` checks: previously,
`modernize` was missing.
* pin go-vet to @v2, remove annoying comment
* add -fips to linux-amd64.tz.gz package
* rm unused setup-chainctl
* use BUILD_TYPE_NAME instead
* mv fips build to enterprise-only
* fixup! use BUILD_TYPE_NAME instead
* temporarily pre-package no plugins for FIPS
* split package-cleanup
* undo package-cleanup, just skip ARM, also test
* skip arm for FIPS in second target too
* fmt Makefile
* Revert "rm unused Dockerfile.fips (for now)"
This reverts commit 601e37e0ff.
* reintroduce Dockerfile.fips and align with existing Dockerfile
* s/IMAGE/BUILD_IMAGE/
* bump the glibc-openssl-fips version
* rm redundant comment
* fix FIPS checks
* set PLUGIN_PACKAGES empty until prepackaged plugins ready
* upgrade glibc-openssl-fips, use non-dev version for final stage
* another BUILD_IMAGE case
* Prepackage the FIPS versions of plugins
* relocate FIPS_ENABLED initialization before use
* s/Config File MD5/Config File Hash/
* Update the FIPS plugin names and encode the + sign
* add /var/tmp for local socket manipulation
---------
Co-authored-by: Alejandro García Montoro <alejandro.garciamontoro@gmail.com>
Co-authored-by: Mattermost Build <build@mattermost.com>
* consolidate go version computation
* use Go from mattermost-build-server
* work around "error obtaining VCS status: exit status 128" in a container"
* no longer require sudo
* fix config-reset for GOFLAGS
* only test-postgres-binary on master/release
* add check-prereqs-enterprise to test-mmctl* to ensure go.work setup
* TestPool
* Store infra
* Store tests updates
* Bump maximum concurrent postgres connections
* More infra
* channels/jobs
* channels/app
* channels/api4
* Protect i18n from concurrent access
* Replace some use of os.Setenv
* Remove debug
* Lint fixes
* Fix more linting
* Fix test
* Remove use of Setenv in drafts tests
* Fix flaky TestWebHubCloseConnOnDBFail
* Fix merge
* [MM-62408] Add CI job to generate test coverage (#30284)
* Add CI job to generate test coverage
* Remove use of Setenv in drafts tests
* Fix flaky TestWebHubCloseConnOnDBFail
* Fix more Setenv usage
* Fix more potential flakyness
* Remove parallelism from flaky test
* Remove conflicting env var
* Fix
* Disable parallelism
* Test atomic covermode
* Disable parallelism
* Enable parallelism
* Add upload coverage step
* Fix codecov.yml
* Add codecov.yml
* Remove redundant workspace field
* Add Parallel() util methods and refactor
* Fix formatting
* More formatting fixes
* Fix reporting
When Elasticsearch/Opensearch was in enterprise repo,
we had to bump up the runners: 9b151defcc.
However, the ES code was move inside server repo,
but the test runners were not changed. This led to frequent
test failures. So we are bumping up the test runners.
This will unfortunately lead to an increased cost, but
we have also cut down in other places viz. the build phase
uses the free runner now (https://github.com/mattermost/mattermost/pull/29297).
And the enterprise build also uses free runner (https://github.com/mattermost/enterprise/pull/1792).
```release-note
NONE
```
Since we have the auto-retry mechanism for replicas,
there is not much benefit in trying 5 times before deciding
to move on.
In the earlier model, we would fail the server startup,
so it made sense to check as many times as possible.
Also reducing the sleep interval so that we can improve
the boot up time in case of a bad replica.
```release-note
NONE
```
* fix: Add job name to the publish report step
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
Co-authored-by: Antonis Stamatiou <stamatiou.antonis@gmail.com>
After moving only contributor PRs to free tier it was
observed that the tests actually finish faster in free tier
than with larger runners. We have let the change soak for about
a month now without any issues.
Therefore we move to free tier runners for all PRs.
```release-note
NONE
```
Co-authored-by: Mattermost Build <build@mattermost.com>
* Constrain when we emit webhooks for retried tests
Only send webhooks (notifying all staff developers) only for retried tests on `master` and `release-*`.
* drop @devs at-mention
---------
Co-authored-by: Mattermost Build <build@mattermost.com>
GitHub actions' YAML supports the `>` directive to a multi-line command to be merged into a single line, but requires that all segments have no leading whitespace otherwise they get silently ignored.
Since this is not nearly as obvious as one might expect, fallback to just the `|` syntax using `\` to extend the lines and keep the overall formatting, fixing the reporting of retried tests via curl webhook.
* disable coverage
This reduces runtime of the server test suite from ~30m to ~10m, and as far as I can see: we discarded the coverage output anyway.
* allow morph 60s to migrate when running tests
* scripts/test.sh: drop COVERMODE
Stop generating coverage data when running unit tests. It's likely we'll want this data back at some point, but for now it's unused and removing simplifies invoking tests for developers.
* scripts/test.sh: remove cleanup steps
* scripts/test.sh: drop TESTS parameter
* scripts/test.sh: drop TESTFLAGS parameter
* switch to gotestsum
* Includes mmctl into the mono-repo
* Update to use the new public module paths
* Adds docs check to the mmctl CI
* Fix public utils import path
* Tidy up modules
* Fix linter
* Update CI tasks to use the new file structure
* Update CI references
We discovered that cancel-in-progress only controls
in-progress jobs. Which means that pending jobs will _always_
be cancelled regardless. There is an open discussion:
https://github.com/orgs/community/discussions/5435
which was closed saying this is how the feature is designed.
We try to work around this by refactoring into separate reusable
workflows and having concurrency only for PR workflows.
```release-note
NONE
```
Co-authored-by: Mattermost Build <build@mattermost.com>
2023-05-09 23:49:06 +05:30
Renamed from .github/workflows/test.yml (Browse further)
