macOS runners getting stuck on Job is about to start running on the hosted runner

[j-bennet]

Describe the bug

The macos jobs on a repo keep getting stuck. From the UI it looks like they are stuck on a "Run tests" step, but there's no output from the tests. The raw log shows this:

2023-05-18T16:47:33.7552355Z Requested labels: macos-latest
2023-05-18T16:47:33.7552694Z Job defined at: coiled/dask-bigquery/.github/workflows/tests.yml@refs/heads/try-fix-ci
2023-05-18T16:47:33.7552805Z Waiting for a runner to pick up this job...
2023-05-18T16:47:34.1259321Z Job is waiting for a hosted runner to come online.
2023-05-18T16:47:40.7586904Z Job is about to start running on the hosted runner: GitHub Actions 6 (hosted)

This is happening on https://github.com/coiled/dask-bigquery.

To Reproduce

Steps to reproduce the behavior:

1. Open a PR into the repo.
2. See the macOS jobs get stuck.

Expected behavior

Jobs finish.

Runner Version and Platform

Version of your runner?

macos-latest

What's not working?

macOS jobs hang while running tests, then timeouts. There is no output from the tests until the job is canceled or times out. There are 3 jobs, sometimes 1 or 2 of them finish successfully.

Job Log Output

2023-05-18T16:47:33.7552355Z Requested labels: macos-latest
2023-05-18T16:47:33.7552694Z Job defined at: coiled/dask-bigquery/.github/workflows/tests.yml@refs/heads/try-fix-ci
2023-05-18T16:47:33.7552805Z Waiting for a runner to pick up this job...
2023-05-18T16:47:34.1259321Z Job is waiting for a hosted runner to come online.
2023-05-18T16:47:40.7586904Z Job is about to start running on the hosted runner: GitHub Actions 6 (hosted)

Runner and Worker's Diagnostic Logs

Log is attached. The job was canceled after producing no output and hanging for 15 or 20 minutes.

stuck-runner-log.txt

[j-bennet]

cc @ncclementi

[ruvceskistefan]

Hey @j-bennet,

I looked into your repo and actions, there is no stuck action on mac os, is this still happening for you or we can close the issue?

[ncclementi]

@ruvceskistefan They all failed because they timed out after a long time, but we can't get consistent runs, for some reason they hang. You can see that the failed ones here https://github.com/coiled/dask-bigquery/actions/runs/5027634794 for example, are all due to time outs.

[jellespijker]

We're experiencing similar behaviour on our repo https://github.com/Ultimaker/cura-binary-data/actions/

[sentrivana]

We're seeing the same for non-MacOS runners with our ubuntu-20.04 workflows.

Some workflows don't even attempt to get a runner and will remain in "Waiting" or "Queued" indefinitely, some do but will get stuck on a variation of:

Requested labels: ubuntu-20.04
Job defined at: getsentry/sentry-python/.github/workflows/test-integration-aiohttp.yml@refs/pull/2181/merge
Waiting for a runner to pick up this job...
Job is waiting for a hosted runner to come online.
Job is about to start running on the hosted runner: GitHub Actions 184 (hosted).

I've now cancelled and rerun the latter and it succeeded, so this is intermittent.

[jacsamell]

We're seeing the same for our builds, gets stuck in Job is about to start:
https://github.com/.../actions/runs/5323119633/jobs/9640530290

Requested labels: ubuntu-latest
Job defined at: .../.github/workflows/root.yml@refs/heads/master
Waiting for a runner to pick up this job...
Job is waiting for a hosted runner to come online.
Job is about to start running on the hosted runner: GitHub Actions 27 (hosted)

Also when try to cancel the job to retry the cancel hangs as well

[FearlessHyena]

If a job hasn't started within a few minutes of Job is about to start running on the hosted runner: *** the runner most likely has issues so it's never going to start
Rather waiting any longer and than timing out, it be better for Actions to automatically try rescheduling the job on another runner at least once before failing
I've also noticed that the job status when queried via the GH API still shows up as queued instead of in_progress even though the runner has already been selected at that point so it's technically not queued anymore

[github-actions]

This issue is stale because it has been open 365 days with no activity. Remove stale label or comment or this will be closed in 15 days.

[github-actions]

This issue was closed because it has been stalled for 15 days with no activity.