starship: Intermittent Test Failures in GH Actions

Bug Report

Our CI appears to be intermittently failing. Last week I got a failure in the go tests which went away on rerun. Just now, we got a failure in nodejs for a commit that only added comments.

I’m not sure what is causing this (a race condition in our test suite, a bug in GH Actions, UB in one of our dependencies) but I’m making an issue to track these. Hopefully we’ll end up seeing some sort of pattern that we can exploit.

If you think you see an intermittent failure in Actions, please post the failed test here along with the following info:

  • Related PR
  • What the change was that triggered the failure
  • The cargo test output for the failed tests.

EDIT: Since transient failures go away when you rerun the tests, don’t be afraid to post the relevant output here before rerunning the tests! You can always delete the comment if the failure sticks around after the rerun.

Solved by https://github.com/starship/starship/issues/942

About this issue

  • Original URL
  • State: open
  • Created 4 years ago
  • Comments: 36 (33 by maintainers)

Most upvoted comments

That particular failure should hopefully no longer happen once the branch is caught up to the changes merged in #867.

I think this was a different issue. I made a PR in #1575 to fix it.

I think it’s possible that the tempdirs get dropped early and deleted.

@andytom Perhaps we’re only getting biased sampling here, but I’m starting to see a pattern: the tests seem to involve filesystem tests (folder_with_* and shows_latest_in_directory_with_*) and most of them are failing by getting None where Some(_) is expected. The one exception to this is the dotnet failure, but without the actual value causing the panic, I can’t say much more there.

Maybe this is a caused by the filesystem not syncing fast enough or reporting that the write has completed when it actually hasn’t. We could try adding some sleeps like we have in git_status testsuite to see if that helps.