Uki: after installation is completed, observed some errors before reboot

[nianyush]

Kairos v3.0.11

[nianyush]

Nothing is broken it seems but it would be nice to get rid of this error

[ci-robbot]

Hello, I'm an experiment by @mudler and @jimmykarily, and I'm here to help with auditing Github issues. I see that your issue titled "Uki: after installation is completed, observed some errors before reboot" (issue #2568) in the "kairos" repository owned by "kairos-io" doesn't seem to have a clear description or steps to reproduce.

Please provide more information about the issue, including a description of the error and steps to reproduce it. Additionally, please mention the versions of the relevant artifacts being used. This will help the community understand your issue better and potentially provide solutions.

Once you provide more information, we'll be happy to assist you further. Remember to follow the project's requirements for opening new issues to ensure it gets the attention it deserves. Thank you!

[jimmykarily]

I think I've even seen a go panic at some point but it was too quick for me to capture it. I'll keep an eye for that to see if it happens again.

[Itxaka]

That seems to be in the cleanup part, which is not a hard failure. Basically the Cleanup object is a stack that we add things that we want to be undone at the end of the action in order.

So we create a dir, add the remove to the Cleanup object, then mount a partition in that dir and put the umount in the Cleanup object. So when the action ends (install, upgrade, reset, etc...) we run those Cleanup functions in parallel.

Usually there is not anything wrong with some of the cleanup functions failing as those are run in the livecd/inmmutable system, they normally wont affect the actual system.

I though that those were not logged unless you were on debug mode though, as we do it on a best-effort capacity. For example, if some of the after-install/install hooks do things to partitions, we cannot anticipate that (i.e. an after install action wrote in the cloud config by the user mounts/unmounts that efi partition, we cannot know in advance) so it should not be a really urgent thing or even a bug.

But indeed, the error message in there is a bit ugly and should not be there.

[Itxaka]

can you trigger it on each run? If so, whats the cloud-config like? Any after-install steps that could be doing something in that efi dir? can you run it with debug mode enabled to see if indeed comes form the Cleanup object (78% sure if comes from that, due to the other logs pointing it to be at the end)

[nianyush]

We don't mount efi dir and the only two things we may mount in after install stage are oem and persistent. Not seeing this every time but it's still quite often

[github-actions[bot]]

This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 5 days.