Upstream Kubernetes Test and Release Overview

• What happens when I submit a PR?
  • Standards enforcement via…
    • Reviews/approvals
    • Verification scripts, e.g. k/kubernetes/hack/make-rules/verify.sh
    • "Presubmit" tests, e.g. sig-node presubmits
• How is the test infrastructure configured?
  • Job definitions: k/test-infra
  • Environment/Infra-as-Code:
    • prow config in k/k8s.io/apps/prow
    • sippy previously deployed in k/k8s.io/apps/sippy@9d11cf23
      • More docs on Sippy, a possible FOSS option for managing flake signal: openshift/sippy
  • Performance: k/perf-tests
    • Performance Dashboard
  • e2e tests overview: k/community
• What happens after a PR is merged?
  • "Postsubmit" tests, e.g. some sig-node periodics
  • Testgrid
    • Overview
    • Individual dashboards, e.g. SIG Release
  • Signal management (release team)
    • Documentation for CI Signal Lead
• Where do the tests live?
  • k/kubernetes
    • k/kubernetes/test/e2e
      • Most upstream e2e tests. Includes most of the 7k e2e tests that need a full cluster to run, including Conformance-tagged tests.
    • k/kubernetes/test/e2e_node
      • Node e2e and conformance tests, that run using a different e2e framework in a single node, limited configuration (e.g. no scheduler). Includes [NodeConformance] tests.
      • See also KEP #3041
  • k/perf-tests
• How do conformance tests work?
  • Documentation
• What does the lifecycle of a release look like?
  • SIG Release has documentation
• How are regressions handled?
  • Cherry-pick documentation
    • example cherry-pick