CPU, RAM and Disk operations are sky high when using self-hosted runners

[sagunkho]

Select Topic Area

Product Feedback

Body

Hey there,
I've been using Github actions for a while now and for the last couple of days when I leave my virtual machines unmonitored, the CPU, RAM and Disk I/O reach 100% and stay there - running the actions-runner on each VM. I've faced this issue with Windows and Linux VMS. I'm not sure what the problem could be or why it happens, it just starts utilizing exactly a 100% of all resources for a long time. I can't work on my host machine due to so many compromised resources. Is it a bug? Is it an exploit? I don't know, only you can tell.

For this reason I've stopped using self-hosted runners. But I would love to start using them again whenever the fix is in place for this high amount of overheating our machines.

Github is awesome by the way, love your work making it simpler for all developers worldwide, its a great opportunity.

[arafajs]

A few things to check:

• Make sure the runner software is up-to-date. GitHub releases periodic updates to the runners to fix bugs and improve performance.
• Check if there are any particular actions or workflows triggering this behavior. It could be something in the code or configuration of those workflows causing the runners to get stuck.
• Look at the runner logs for any errors or warnings that could provide clues. The logs are usually found in the ./actions-runner/ directory.
• Try rebooting the VMs and runners to clear out any stuck processes.
• Consider limiting the number of concurrent jobs allowed on each self-hosted runner to prevent resource exhaustion.
• Switch the runner application to a lower CPU priority via Nice/Renice if on Linux.
• Double check that nothing else on the VMs themselves is using resources, like other unrelated background processes.

[sagunkho]

@arafajs I've double checked the list. There are no issues with the VM or code.

[sagunkho]

[richard-scott]

I have noticed this with buildx running for only the linux/arm/v6 platform on an X64 Github runner. It doesn't matter if it's a self-hosted runner or a Github runner, they both still consume 100% of all CPUs.