Extract streaming code into dedicated staging modules while keeping stable compatibility APIs for external client-go consumers. This commit: - adds `k8s.io/cri-streaming` for CRI exec/attach/portforward server code - adds `k8s.io/streaming` as the canonical home for shared transport primitives (`httpstream`, `spdy`, `wsstream`, runtime helpers) - switches in-tree transport consumers to `k8s.io/streaming` - removes in-tree kubelet CRI streaming package - preserves NO_PROXY/no_proxy CIDR handling in extracted SPDY proxier logic - adds deprecated `k8s.io/apimachinery/pkg/util/httpstream` compatibility wrappers (`httpstream`, `spdy`, `wsstream`) backed by `k8s.io/streaming` - restores exported client-go SPDY/portforward API signatures to apimachinery `httpstream` types for downstream compatibility - adds streaming-native client-go adapters/constructors so in-tree callers can use `k8s.io/streaming` without changing external compatibility APIs - deduplicates SPDY-over-websocket dial negotiation shared by compat and streaming tunneling dialers - logs dropped unknown stream types in `RemoveStreams` adapter fallbacks to improve compatibility-path debuggability - adds integration coverage for the streaming-upgrader-to-client-go-compat adapter path against a real cri-streaming exec endpoint - clarifies kubectl streaming import aliasing to avoid `httpstream` package ambiguity - updates tests, import restrictions, publishing metadata, and vendor/module metadata for the new staging modules Signed-off-by: Davanum Srinivas <davanum@gmail.com> |
||
|---|---|---|
| .. | ||
| auth | ||
| autoscaling | ||
| config | ||
| conformance | ||
| daemonset | ||
| debug | ||
| deployment | ||
| endpointslice | ||
| events | ||
| gpu | ||
| internal | ||
| job | ||
| kubectl | ||
| kubelet | ||
| kubesystem | ||
| manifest | ||
| metrics | ||
| network | ||
| node | ||
| perf | ||
| pod | ||
| providers | ||
| pv | ||
| rc | ||
| registry | ||
| replicaset | ||
| resource | ||
| security | ||
| service | ||
| skipper | ||
| ssh | ||
| statefulset | ||
| testfiles | ||
| timer | ||
| volume | ||
| websocket | ||
| .import-restrictions | ||
| bugs.go | ||
| expect.go | ||
| expect_test.go | ||
| flake_reporting_util.go | ||
| framework.go | ||
| get.go | ||
| ginkgologger.go | ||
| ginkgowrapper.go | ||
| ginkgowrapper_test.go | ||
| gomega.go | ||
| log.go | ||
| log_test.go | ||
| namespacedname.go | ||
| nodes_util.go | ||
| OWNERS | ||
| ports.go | ||
| provider.go | ||
| README.md | ||
| size.go | ||
| test_context.go | ||
| timeouts.go | ||
| util.go | ||
Overview
The Kubernetes E2E framework simplifies writing Ginkgo tests suites. It's main usage is for these tests suites in the Kubernetes repository itself:
- test/e2e: runs as client for a Kubernetes cluster. The e2e.test binary is used for conformance testing.
- test/e2e_node: runs on the same node as a kubelet instance. Used for testing kubelet.
- test/e2e_kubeadm: test suite for kubeadm.
Usage of the framework outside of Kubernetes is possible, but not encouraged. Downstream users have to be prepared to deal with API changes.
Code Organization
The core framework is the k8s.io/kubernetes/test/e2e/framework package. It
contains functionality that all E2E suites are expected to need:
- connecting to the apiserver
- managing per-test namespaces
- logging (
Logf) - failure handling (
Fail,Failf) - writing concise JUnit test results
It also contains a TestContext with settings that can be controlled via
command line flags. For historic reasons, this also contains settings for
individual tests or packages that are not part of the core framework.
Optional functionality is placed in sub packages like
test/e2e/framework/pod. The core framework does not depend on those. Sub
packages may depend on the core framework.
The advantages of splitting the code like this are:
- leaner go doc packages by grouping related functions together
- not forcing all E2E suites to import all functionality
- avoiding import cycles
Execution Flow
When a test suite gets invoked, the top-level Describe calls register the
callbacks that define individual tests, but does not invoke them yet. After
that init phase, command line flags are parsed and the Describe callbacks are
invoked. Those then define the actual tests for the test suite. Command line
flags can be used to influence the test definitions.
Now Context/BeforeEach/AfterEach/It define code that will be called later
when executing a specific test. During this setup phase, f := framework.NewDefaultFramework("some tests") creates a Framework instance for
one or more tests. NewDefaultFramework initializes that instance anew for
each test with a BeforeEach callback. Starting with Kubernetes 1.26, that
instance gets cleaned up after all other code for a test has been invoked, so
the following code is correct:
f := framework.NewDefaultFramework("some tests")
ginkgo.AfterEach(func() {
# Do something with f.ClientSet.
}
ginkgo.It("test something", func(ctx context.Context) {
# The actual test.
})
Optional functionality can be injected into each test by adding a callback to
NewFrameworkExtensions in an init function. NewDefaultFramework will invoke
those callbacks as if the corresponding code had been added to each test like this:
f := framework.NewDefaultFramework("some tests")
optional.SomeCallback(f)
SomeCallback then can register additional BeforeEach or AfterEach
callbacks that use the test's Framework instance.
When a test runs, callbacks defined for it with BeforeEach and AfterEach
are called in first-in-first-out order. Since the migration to ginkgo v2 in
Kubernetes 1.25, the AfterEach callback is called also when there has been a
test failure. This can be used to run cleanup code for a test
reliably. However,
ginkgo.DeferCleanup
is often a better alternative. Its callbacks are executed in first-in-last-out
order.
test/e2e/framework/internal/unittests/cleanup/cleanup.go shows how these
different callbacks can be used and in which order they are going to run.