alexcrichton opened PR #11039 from alexcrichton:remove-cancel-on-failure to bytecodealliance:main:
The original intent of this is that as soon as a job fails in the merge queue it should get itself out of the merge queue as soon as possible. That requires cancelling all other active jobs because the way we have things set up is that there's a "join node" at the end of CI which collects the status of all other jobs and reflects them. This is what the merge queue then listens for. By default that means that when a job fails the CI build will continue until everything else finishes, and only then will the PR be removed from the merge queue.
The main drawback of this cancel-on-failure logic is that it seems to race with how github actions classifies a job. This is a situation where a job cancels itself and so there's a race between the cancellation cancelling the job vs the job finishing and flagging its failed status. That means that jobs can exit the merge queue and have no apparent failures which requires digging through the summary of all jobs to figure out which one actually failed. Personally I find this annoying and obtuse and the motivation for this commit is to solve this.
The downside of this change is that if a job fails in the merge queue it'll block the merge queue for longer than it otherwise would. Given that our merge queue is not constantly full, however, I'm assuming that this is generally ok.
<!--
Please make sure you include the following information:
If this work has been discussed elsewhere, please include a link to that
conversation. If it was discussed in an issue, just mention "issue #...".Explain why this change is needed. If the details are in an issue already,
this can be brief.Our development process is documented in the Wasmtime book:
https://docs.wasmtime.dev/contributing-development-process.htmlPlease ensure all communication follows the code of conduct:
https://github.com/bytecodealliance/wasmtime/blob/main/CODE_OF_CONDUCT.md
-->
alexcrichton requested dicej for a review on PR #11039.
alexcrichton requested wasmtime-default-reviewers for a review on PR #11039.
dicej submitted PR review.
alexcrichton merged PR #11039.
Last updated: Dec 06 2025 at 07:03 UTC