Skip to content

Latest commit

 

History

History
82 lines (47 loc) · 3.66 KB

README.md

File metadata and controls

82 lines (47 loc) · 3.66 KB
Raw

How to work with Jenkins jobs

Each job has:

  1. A Jenkins job entry, in either kubernetes-jenkins or kubernetes-jenkins-pull.

  2. A json entry in job configs

  3. (Required for e2e jobs) a foo.env match job foo in /jobs

  4. (Required for ci jobs) a testgrid config entry

Ping @kubernetes/test-infra-maintainers if you have any questions.

Add a Jenkins job

Say you want to add a new job, foo.

For CI jobs

  1. CI job configs are located under job-configs/kubernetes-jenkins.

    For e2e jobs, add an entry to bootstrap-ci.yaml.

    For PR jobs, add an entry to bootstrap-pull.yaml.

    For jobs that trigger on a merge (such as a build job), add an entry to bootstrap-ci-commit.yaml.

    For jobs that clone a repo, add an entry to bootstrap-ci-repo.yaml.

    For soak jobs, add an entry to bootstrap-ci-soak.yaml.

    For maintenance jobs, add an entry to bootstrap-maintenance-ci.yaml.

    Run ./jenkins/bootstrap.py --help will help you determine which flag you might need.

  2. Add an entry to config.json. Choose an appropriate scenario file and args.

  3. If it's an e2e job, add foo.env to jobs, which defines environment variables your job will be using. You can reference it from other e2e jobs.

  4. Add your new job to test-infra/testgrid/config/config.yaml, instruction can be found in here.

  5. Make sure all presubmit tests pass. Running bazel test //... locally is an quick way to trigger most of the unit tests.

For PR jobs

  1. PR job configs are under job-configs/kubernetes-jenkins-pull.

    You are mostly want to add your PR job entry to bootstrap-pull.yaml.

  2. PR jobs are triggered by prow, add your entry to prow config as well.

  3. Make sure all presubmit tests pass. Running bazel test //... locally is an quick way to trigger most of the unit tests.

Run a Jenkins job locally

Below paths are from the root of the kubernetes/test-infra repo.

The straight forward way to mimic a job run on Jenkins is, for example:

./jenkins/bootstrap.py --job=ci-kubernetes-e2e-gce-canary --json=1 --bare --timeout=40

If you want to upload the log and display in gubernator, you can append

--service-account=/path/to/service-account-cred.json --upload='gs://your-bucket-name/logs'

Reference: Creating and Enabling Service Accounts for Instances

You can also specify --repo, --branch, and other bootstrap flags, details can be found by running ./jenkins/bootstrap.py --help

If you want to overwrite some scenario flags, you can change config.json locally.

Alternatively, if you do not care about logs and artifacts, you can also run:

export BOOTSTRAP_MIGRATION=true # https://github.com/kubernetes/test-infra/pull/1801#discussion_r102347949
export BUILD_NUMBER=SOMETHING
export JOB_NUMBER=JOB_NAME
./scenario/your_scenario.py --env-file=your_env_file --scenario-flag=your-flag