Startup Failure status displaying as queued

[obrienrobert]

Hello! I noticed that the following metric and selectorgithub_workflow_status{status="queued"} lists runs that are in a status of Startup Failure also.

I would expect to see workflow runs that are queued and not in a Startup Failure state, although looking at the REST API it seems that Startup Failure isn't actually a status (maybe it's displaying the previous known status?)

Can be one of: completed, action_required, cancelled, failure, neutral, skipped, stale, success, timed_out, in_progress, queued, requested, waiting, pending

[tboerger]

Not sure about that... So far I have no example to compare that properly. We are just consuming everything 1 to 1 what GitHub gives us.

[AppliNH]

Hi! In the meantime you can use the following prometheus expression to filter out these ocurrences

github_workflow_run_duration_run_created_minutes{status=~"queued|pending"} 
* on (run) 
group_left()
(
  label_replace(
    (max by (run_id) (github_workflow_job_status == 1)),
    "run",
    "$1",
    "run_id",
    "(.+)"
  )
)

[tboerger]

I think this should be resolved with v4.1.0, if the issue still exists we could reopen it.

[AppliNH]

Hello! We just updated to 4.1.1 and the issue is still there. Can you point to the commit that's supposed to solve this?

[tboerger]

I can't remember the exact commit, but there had been changes related to the updated fields in the database like conclusion and status correctly updating.

[tboerger]

BTW, the release of 4.1.0 is already more than 4 months old.

[AppliNH]

Yes, we've been slow with the upgrades as we didn't need them yet. Also, we didn't take time to properly review the breaking changes hence this preliminary upgrade to 4.1.1 👍

We'll upgrade to the latest release, and I'll let you know via a new ticket if the issue is still there! We're also investigating on our side.

Thanks 🙏