Called workflows cannot be queued onto self-hosted runners across organisations/enterprises. Failed to queue this job. Labels: 'ubuntu-latest'

Reusable workflows being called from another repository are failing with the error in the title.

This is the same problem as: 'ubuntu-latest' runner not found when using workflow_call between repositories - #16 by mohamed3114
but that has already been marked solved, and this is a new issue or some kind of regression, because it only just started happening.

There seems to be something to this reply, which mentions that this only happens when the concurrency limit is reached. When I retry the jobs with available capacity, they work, but every time there’s contention, these jobs fail now.

It’s very disruptive and means I need to babysit my jobs so that I can re-submit the ones that use reusable workflows from another repository, once I can see that they won’t be queued. And in the meantime, I am getting emailed about these failed jobs too.

I should add, there are NO self-hosted runners in use in my repository.

cc @jenschelkopf (since you resolved the previous issue)

4 Likes

Can anyone help with this? It’s so disruptive, every PR, commit, merge has these jobs that originate from reusable workflows failing and forcing me to watch and wait until there’s capacity before I re-run them…

Well this is my “me-too” post. You are absolutely right - it seems when a large number (~15-25 concurrent jobs) for a specific repo, this pops up.

You are absolutely right that is very disruptive that my production pipeline fails because 2 people just pushed to feature branches.

So much for velocity!

1 Like

Don’t have much to add other than this is holding up a nifty new feature for our repos as well.

1 Like

Adding my voice to the chorus here. Any idea when this bug will be addressed?

1 Like

Thanks for reporting this; the team is investigating. I hope to have more to share soon.

2 Likes

Great to hear! Looking forward to a fix… :pray:

3 days later and still no response.
Finally got our workflows up and “accepted” by the GitHub Actions for our pipeline and getting the same error.

We have a Self-Hosted agent running with the tags required but GitHub actions is refusing to pick up and run it on the runner.

This exact flow was working perfectly fine on a personal / private repo with exactly the same agent and machine. Only thing that has changed is that it is now an Organisation runner.

Ok, re-read the original report and it notes that they are not using self-hosted runners, my bad. But issue persists

Yeah based on the error message it seems known/intentional that cross-org on self-hosted runners is not supported (though I have no idea why, that doesn’t seem intuitive at all); however this thread/topic is mainly to report that this error is newly, erroneously returned on GitHub runners. I recommend finding or starting a new topic to request a change in the functionality for cross-org self-hosted runners.

I have here :smiley:

Called workflows cannot be queued onto self-hosted runners across organisations/enterprises. Failed to queue this job. Labels: ‘windows’ - Code to Cloud / GitHub Actions - GitHub Community

@ethomson any update?
Does this warrant an issue being created in the runner repository?

Or is there any other way I can track progress on it?

@ethomson please can someone help? The problem is not only still occurring, it’s getting way worse. The jobs are now failing even more quickly, even when there are no other workflows running. They sit for a little over 1 minute and then die with the false self-hosted runner message, and all I can do is to keep trying over and over again and hope it runs.
This run, which has 3 jobs that run successively not in parallel, with nothing else running at all, took 11 attempts before I could finally run it!

This experience is really awful. It’s killing my velocity and motivation to work on anything. Please help!!

@ethomson please, is there any update?

Hi Brian - I don’t have an ETA but I checked in with the engineering team and they believe they have identified the problem and occasions on which this occurs. They should be working on a fix now. Apologies for the delay.

1 Like

Thanks @ethomson , did something change yesterday/today? I’m seeing one of these jobs properly queue without failing, so that’s hopeful… :pray:

Yes, the fixes for this should be completely rolled out to all customers now. Thanks for letting us know that it’s working and thanks again for your patience.

1 Like

Hi ethomson , seems like this is still occurring I’ve been facing it all day , here is my .yml file :

name: Super-Linter

on: push

jobs:
  super-lint:
    name: Lint code base
    runs-on: self-hosted
    steps:
      - name: Checkout code
        uses: actions/checkout@v2

      - name: Run Super-Linter
        uses: github/super-linter@v3
        env:
          DEFAULT_BRANCH: main
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

This looks it might be a different issue. This thread is specifically about reusable workflows, your workflow is not a definition of a resuable workflow, and it is not consuming one either. You may be better served posting a new thread for this one.