If you are using a released version of Kubernetes, you should refer to the docs that go with that version.
Documentation for other releases can be found at releases.k8s.io.
This assumes you already read the development guide to install go, godeps, and configure your git client.
In order to send pull requests you need to make sure you changes pass unit and integration tests.
Kubernetes only merges pull requests when e2e tests are passing, so it is often a good idea to make sure these work as well.
- Unit tests should be fully hermetic
- Only access resources in the test binary.
- All packages and any significant files require unit tests.
- The preferred method of testing multiple scenarios or inputs
is table driven testing
- Example: TestNamespaceAuthorization
- Unit tests must pass on OS X and Windows platforms.
- Tests using linux-specific features must be skipped or compiled out.
- Skipped is better, compiled out is required when it won't compile.
- Concurrent unit test runs must pass.
- See coding conventions.
cd kubernetes
hack/test-go.sh # Run all unit tests.
cd kubernetes
# Run all tests under pkg (requires client to be in $GOPATH/src/k8s.io)
godep go test ./pkg/...
# Run all tests in the pkg/api (but not subpackages)
godep go test ./pkg/api
Running the same tests repeatedly is one way to root out flakes. You can do this efficiently.
cd kubernetes
# Have 2 workers run all tests 5 times each (10 total iterations).
hack/test-go.sh -p 2 -i 5
For more advanced ideas please see flaky-tests.md.
Currently, collecting coverage is only supported for the Go unit tests.
To run all unit tests and generate an HTML coverage report, run the following:
cd kubernetes
KUBE_COVER=y hack/test-go.sh
At the end of the run, an the HTML report will be generated with the path printed to stdout.
To run tests and collect coverage in only one package, pass its relative path under the kubernetes
directory as an argument, for example:
cd kubernetes
KUBE_COVER=y hack/test-go.sh pkg/kubectl
Multiple arguments can be passed, in which case the coverage results will be combined for all tests run.
Coverage results for the project can also be viewed on Coveralls, and are continuously updated as commits are merged. Additionally, all pull requests which spawn a Travis build will report unit test coverage results to Coveralls. Coverage reports from before the Kubernetes Github organization was created can be found here.
To run benchmark tests, you'll typically use something like:
cd kubernetes
godep go test ./pkg/apiserver -benchmem -run=XXX -bench=BenchmarkWatch
This will do the following:
-run=XXX
will turn off regular unit tests
- Technically it will run test methods with XXX in the name.
-bench=BenchmarkWatch
will run test methods with BenchmarkWatch in the name
- See
grep -nr BenchmarkWatch .
for examples
-benchmem
enables memory allocation stats
See go help test
and go help testflag
for additional info.
- Integration tests should only access other resources on the local machine
- Most commonly etcd or a service listening on localhost.
- All significant features require integration tests.
- This includes kubectl commands
- The preferred method of testing multiple scenarios or inputs
is table driven testing
- Example: TestNamespaceAuthorization
- Integration tests must run in parallel
- Each test should create its own master, httpserver and config.
- Example: TestPodUpdateActiveDeadlineSeconds
- See coding conventions.
Kubernetes integration tests require your PATH to include an etcd installation. Kubernetes includes a script to help install etcd on your machine.
# Install etcd and add to PATH
# Option a) install inside kubernetes root
cd kubernetes
hack/install-etcd.sh # Installs in ./third_party/etcd
echo export PATH="$PATH:$(pwd)/third_party/etcd" >> .profile # Add to PATH
# Option b) install manually
cd kubernetes
grep -E "image.*etcd" cluster/saltbase/etcd/etcd.manifest # Find version
# Install that version using yum/apt-get/etc
echo export PATH="$PATH:<LOCATION>" >> .profile # Add to PATH
cd kubernetes
hack/test-integration.sh # Run all integration tests.
- e2e tests build kubernetes and deploy a cluster of nodes.
- Generally on a specific cloud provider.
- Access gcr.io images
- Access a specific, non-latest image tag (unless testing pulling).
- Tests may not flake due to intermittent issues.
- Use ginko to desribe steps.
- Use NewDefaultFramework
- Contains clients, namespace and auto resource cleanup
- See coding conventions.
In general passing unit and integration tests should provide sufficient confidence to allow code to merge. If that is not the case, please invest more time adding unit and integration test coverage. These tests run faster and have a smaller failure domain.
However, end-to-end (e2e) tests provide maximum confidence that the system is working in exchange for reduced performance and a higher debugging cost.
e2e tests deploy a real kubernetes cluster of real nodes on a concrete provider such as GCE. The tests then manipulate the cluster in certain ways and assert the expected results.
For a more in depth discussion please read End-to-End Testing in Kubernetes.
cd kubernetes
go run hack/e2e.go -v --build --up --test --down
# Change code, run unit and integration tests
# Push to an existing cluster, or bring up a cluster if it's down.
go run hack/e2e.go -v --pushup
# Run all tests on an already up cluster
go run hack/e2e.go -v --test
# Run only conformance tests
go run hack/e2e.go -v -test --test_args="--ginkgo.focus=\[Conformance\]"
# Run tests on a specific provider
KUBERNETES_PROVIDER=aws go run hack/e2e.go --build --pushup --test --down
For a more in depth discussion please read End-to-End Testing in Kubernetes.