[ws-daemon] fails mounting proc

[svenefftinge]

There are regular CRITICAL errors logged from the nsinsider process. Followed by a "cannot mount /proc" error logging.

I am not sure how this is experienced by users, but I guess it means a workspace start failure.

[csweichel]

Unfortunately the error message isn't very helpful. It's well possible that the mount-point where the user is trying to mount proc no longer exists. It's also just as possible that this is an actual problem.

First step is to get better logging in place.

[csweichel]

I've raised a PR to improve the logs in this corner of the system. We'll deploy today 13:00 CEST and bring this into prod.
We'll revisit this issue once the logging is in place.

[csweichel]

looking at the non-headless cases it seems that the proc mount failures are due to Docker container starts going wrong.
I'd propose we tone down this particular message (i.e. the mount point disappears before we can mount/does not exist).
If this affected workspaces, we'd see workspaces fail.

That said, it would be interesting to reproduce the issue and find out how it comes to be.

For now, I'll reduce the priority because this does not seem to directly impact users.

[csweichel]

/schedule

[csweichel]

Closing for now - will reopen if the issue persists