The CI task fail just because VM kill fail

[montjoie]

Hello

I do some tests in openBSD and it fail with:

Tearing down VM
Syncing back files
/usr/bin/rsync -auz runner@cross_platform_actions_host:/home/runner/work/ /home/runner/work
Terminating VM
/usr/bin/sudo kill -s TERM 2225
kill: (2225): No such process
Error: The process '/usr/bin/ssh' failed with exit code 1

Since everything in my CI is okay, it is sad that the task is still set as BAD due to internal errors.

[jacob-carlborg]

Could you please link to the failing job?

[montjoie]

https://github.com/montjoie/xymon-upstream/actions/runs/21206914637/job/61005520014

[jacob-carlborg]

The two lines from your output are expected:

/usr/bin/sudo kill -s TERM 2225
kill: (2225): No such process

They always occur, even for successful jobs. Have a look at one of our successful jobs [1]. First the action tries to gracefully shutdown the VM. Sometimes that doesn't work. Then it tries to forcefully terminate the VM. But instead of looking if the VM process is still running, it just tries to terminate it and ignoring any errors. If it would first check if the VM process is running, that would cause a race condition. If it fails to terminate the process there's not much the action can do anyway. If it fails to terminate the process because it doesn't exist, that's fine because it should be terminated anyway. Or it would successfully terminate the process because it previously failed to gracefully shut it down.

It looks like the last three commands of your job can fail and would not print anything if they failed [2]. I would start and debug those to see if any of those commands are failing.

[1] https://github.com/montjoie/xymon-upstream/actions/runs/21206914637/job/61005520064#step:3:498
[2] https://github.com/montjoie/xymon-upstream/actions/runs/21206914637/job/61005520014#step:3:103