Skip to content

ci: test flaky CI on clean main#2938

Draft
wen-coding wants to merge 5 commits intomainfrom
wen/test_flaky_ci
Draft

ci: test flaky CI on clean main#2938
wen-coding wants to merge 5 commits intomainfrom
wen/test_flaky_ci

Conversation

@wen-coding
Copy link
Contributor

Summary

Empty commit on unmodified main to verify whether these CI tests are pre-existing flaky:

  • TestTimelyProposal
  • TestAvailClientServer
  • TestStateLock_NoPOL
  • TestStateLock_POLSafety1

Draft PR — do not merge.

Made with Cursor

@github-actions
Copy link

github-actions bot commented Feb 19, 2026

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

BuildFormatLintBreakingUpdated (UTC)
✅ passed✅ passed✅ passed✅ passedFeb 20, 2026, 2:44 PM

wen-coding and others added 4 commits February 19, 2026 20:15
Empty commit to trigger CI and check if TestTimelyProposal,
TestAvailClientServer, TestStateLock_NoPOL, and
TestStateLock_POLSafety1 fail on unmodified main.

Co-authored-by: Cursor <cursoragent@cursor.com>
- pbts_test.go: bump timeoutPropose from 10-50ms to 500ms so proposals
  arrive before the timeout fires under CI load
- syncer_test.go: bump sleep from 50ms to 500ms so the goroutine has
  time to process chunk refetch under load
- node_test.go: bump context timeout from 10s to 60s so the node has
  time to produce its first block under load

Co-authored-by: Cursor <cursoragent@cursor.com>
Many consensus tests (e.g. TestStateLock_POLSafety1, TestStateLock_NoPOL)
race between SetProposalAndBlock and the propose timeout. Under CI load,
40ms is not enough margin for the proposal to be delivered before the
timeout fires, causing the node to prevote nil and the test to fail.

Increasing to 500ms gives ample margin while keeping test runtime
acceptable (~500ms per propose-timeout wait step).

Co-authored-by: Cursor <cursoragent@cursor.com>
500ms was overly generous and could contribute to slow test suites under
CI. 200ms still gives 5x margin over the original 40ms, which is enough
to prevent the SetProposalAndBlock race from losing to the timeout.

Co-authored-by: Cursor <cursoragent@cursor.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant