GitHub runner randomly kills process

I’m trying to build a project through gradle on GitHub actions which delegates to a bash script which is supposed to compile GCC. At some point during the run the (I would asssume) github runner decides to SIGKILL the process.

The runs in question can be found here: https://github.com/supergrecko/gccjit/pull/1

If you inspect the logs from the job, you’ll see something like this, signaling that the gradle command was stopped through SIGKILL from the bash script in /home/runner/work/_temp/{{ some uuid }}.sh

Previously I would run Gradle through its daemon, but it seems the runner would end it while it was performing work.

(I would embed another picture here, but the forum limits me to one media embed. You can find it at the end of the log of this run)

I would assume the process is SIGKILL’ed because the system assumes it’s inactive because it has delegated all it’s work to a subprocess. Is there anything I can do to prevent the runner from ending my gradle process?

@supergrecko,

At some point during the run the (I would asssume) github runner decides to SIGKILL the process.

You used the GitHub-hosted runners to run the jobs, right?
If so, please check with the following steps:

  1. Try using the self-hosted runners that are installed on your machines (local or VM) to see if the same issue occurs.
  2. Directly execute the script on your machines to see if the same issue occurs.

If the issue occurs only when you execute in the workflow, there may have some problems for the runners. You can go to the “actions/runner” repository, and report an issue to the appropriate engineering team for further investigation and evaluation.

In addition, we seem can’t access the link of the repository you shared above. It is private, and we can’t access. You can share the example on a public repository.