Skip to content

Improve e2e lifecycle cleanup and PR workflow - #63

Open
bittermandel wants to merge 1 commit into
mainfrom
chore/e2e-cleanup-pr
Open

Improve e2e lifecycle cleanup and PR workflow#63
bittermandel wants to merge 1 commit into
mainfrom
chore/e2e-cleanup-pr

Conversation

@bittermandel

Copy link
Copy Markdown
Contributor

No description provided.

bittermandel added a commit that referenced this pull request Apr 27, 2026
)

* ci(e2e): cancel concurrent runs and bound job timeout

Adds a concurrency group keyed on workflow + ref so a new push to a PR
cancels the in-flight run instead of queuing behind it. Restricts push
triggers to main, expands PR triggers (opened/synchronize/reopened/
ready_for_review), and caps job runtime at 90 minutes so a wedged Kind
or hung process can't burn an Actions runner indefinitely.

* test(e2e): run cleanup on failure and use baseline namespace policy

Three small robustness fixes to the e2e harness, ported from PR #63:

- make test-e2e wraps the Go test invocation in trap cleanup EXIT so
  Kind teardown runs even when tests fail (previously the cleanup line
  was only reached on success).
- The namespace is created via kubectl apply --dry-run=client | apply
  instead of kubectl create, so re-running the suite against an existing
  cluster doesn't fail with AlreadyExists.
- Namespace policy switched from restricted to baseline. Neon pods need
  baseline (runAsUser=1000 etc.); restricted blocks them silently.
- Teardown calls (pod, namespace) gain --ignore-not-found and --wait=false
  so partial failures don't stall AfterAll. make undeploy/uninstall pass
  ignore-not-found=true and undeploy is bounded with a 2-minute context.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant