Jobs are randomly skiped or cancelled with unknown errors

[njzjz]

Select Topic Area

Bug

Body

Starting yesterday, I have seen a lot of jobs randomly skipped or cancelled with unknown errors.

Example: https://github.com/deepmodeling/deepmd-kit/actions/runs/9293232817

[indiarasabarreto]

Because if it's not a failure and not cancelled it's either success() or skipped.

1. if: github.event_name == 'push' && !failure() && !cancelled()
2. Same thing. Add the if: condition to make sure it would behave the same as the previous step, even though there might exist some sort of inheritance I didn't dig through.

[saltydk]

This is an issue with GHA, not his conditionals, it cancelled the build without anything happening inside of the run.

[VietND96]

+1, the same for me in #126539

[grandmas-favorite]

Hi everyone (@njzjz, @VietND96, @saltydk, and @indiarasabarreto)!

We had found an issue where small set of jobs are being canceled while they are in progress, but it should have been resolved a couple of hours ago.

Cheers!

[hanabi1224]

+1 I just experienced the same. Jobs with conditions if: github.event.pull_request.draft == false get skipped when the PR is no longer a draft. Retry does not fix the issue and I have to convert the PR to draft and then back to make it work.

https://github.com/ChainSafe/forest/actions/runs/9367294602/job/25786761395?pr=4387

P.S. When all CI jobs pass in ChainSafe/forest#4387 I see 20 CI job failures on my iOS Github app, refreshing does not help

[saltydk]

I doubt that is related to this issue.