update-check in Manage Integration Check workflow sometime fails
#8937
Comments
|
https://github.com/Agoric/agoric-sdk/actions/runs/8621724430 Is this failed run something to be concerned with? |
|
yeah it's the race i've seen before. usually re-running will fix |
mergify bot
added a commit
that referenced
this issue
Aug 2, 2024
…low (#9787) refs: #8937 ## Description Addresses the issue with flake in the `manage-integration-check.yml` workflow. This is due to a race condition where the workflow is executed 3 times in quick succession and is expected to run in order of execution. The "Integration tests" workflow triggers `manage-integration-check.yml` 3 times. once for `requested`, `in_progress`, and `completed` statuses of the calling workflow. In cases where the `completed` status might happen at the same time as the other two (such as when the "Integration tests" workflow is skipped), the 3 calls of the `manage-integration-check.yml` workflow might run simultaneously and cause a race condition. the race condition occurs because the final call of `manage-integration-check.yml` (the one triggered by the `completed` status of "Integration tests") expects a github check run to already exist. since this might not be the case in a race condition, it fails To fix this two things have been done: - instead of crashing, the job now creates a new github check run with a `completed` status - since the previous two jobs will run after the final job, they might accidentally overwrite the github check run with the `completed` status and accidentally create a check run with `in_progess` status. their filter check has been removed entirely so that they consider `completed` check runs as well. ### Security Considerations ### Scaling Considerations ### Documentation Considerations ### Testing Considerations ### Upgrade Considerations
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the bug
The job sometimes fail, not sure why? I'm positing a sort of race where a previous
create-checkdidn't complete yet.Arguably the
update-checkshould create if it doesn't exist, or retry?The consequence is pretty bad: the
integration-test-resultcheck never completes, and the PR gets stuck.To Reproduce
Expected behavior
Completion result of integration test workflow always fully reflected in check result.
Platform Environment
CI
Context
#8731
Screenshots
The text was updated successfully, but these errors were encountered: