How can I clear a stale job/check from an already completed Actions run?

[hoons1994]

🏷️ Discussion Type

Question

💬 Feature/Topic Area

Other

Discussion Details

Problem

An API 37 job from the first attempt of a GitHub Actions workflow still reports in_progress, although all of its steps (including the full instrumented test suite and Complete job) completed successfully. The parent workflow run is already completed with success.

• Repository: https://github.com/hoons1994/ShutterSoundZero
• Workflow run: https://github.com/hoons1994/ShutterSoundZero/actions/runs/40511574913
• Stale first-attempt job/check: https://github.com/hoons1994/ShutterSoundZero/actions/runs/40511574913/job/122214660416
• Commit: db7d3624f437eee22a63982829e173bdf8c262ab

The jobs and check-runs APIs return status: in_progress, conclusion: null, and completed_at: null for the first attempt, even though its individual steps show success.

What I tried

I reran only the API 37 job. The second attempt completed successfully, and its job and check are both completed/success: https://github.com/hoons1994/ShutterSoundZero/actions/runs/40511574913/job/122223698249. The first-attempt job remains in_progress. GitHub Support closed my request and referred me to self-service resources.

I found https://github.com/orgs/community/discussions/209225, but in that case the parent workflow is also stuck; here the parent workflow has already completed successfully.

Is there a supported way to clear or correct the stale status of only the first attempt while preserving the workflow history and the successful second attempt? I would prefer not to delete the workflow run. Thanks for any guidance.

[Sahil-dev7]

You cannot fix it by yourself; GitHub must repair the stale record if a backend correction is available.

[kittimzhe]

Live update: that job (122214660416) now reports completed / success — started Oct 27 13:23 UTC, completed 13:32, still run_attempt: 1. The in_progress state you saw has resolved itself.

What you hit is the job-state lag variant of the current slow-processing window: steps finish, but the job-level status event lands late — same family as the run-level lags documented in #209225 (where a couple of runs self-healed in ~75 minutes). Yours took longer but followed the same curve: nothing was actually stuck, the completion event was just queued behind the backlog.

For future occurrences, the practical levers while waiting:

• If a stale check is blocking a PR merge, pushing an empty commit regenerates the check suite fresh (usually clears the block before the lag catches up).
• If it's a required check pinned to the stale job, branch protection's re-run link or an admin merge covers the gap.
• Genuine multi-day in_progress with zero steps progressing is the different variant (runner-side wedge) — that one justifies a support ticket with the job URL.

[hoons1994]

Thank you for checking and for the guidance! I confirmed via the GitHub Actions jobs and check-runs APIs that the first attempt now reports completed / success, and the second attempt remains successful as well. The stale status has cleared, so the issue is resolved. I appreciate your help!