Attempting to stop frozen instances causes unintended behaviour

Incus
1

When an instance is frozen attempting to stop the instance causes Incus to get stuck, incus list shows the instance as Running, but any attempt to change the state of the instance fails, however, it is still available, pings from other instances on the same network succeed and processes continue running.
attempting to restart Incus fails as the stop process is ongoing, even attempting to shut down the host machine fails.

When I first discovered this issue I rebooted the host machine, and after some time waiting for the process to be killed It was fixed. The second time while testing I left the machine running, after a long time (2-3 hours) the issue resolved itself.

I have read the docs but haven’t seen anything related to this issue, is this intended?, or is this a bug of some sort?

2

What version of Incus and the Linux kernel are you running?

Incus itself failing to start back sounds a lot like a kernel bug of some kind.
If it was just incus stop not working on a paused instance, that’s somewhat common. We’ve made some tweaks to make that more likely to work, but there are still cases that will need an incus stop -f.

A stuck incus stop will hold a systemctl restart incus for the default max wait time of 5 minutes but not longer than that, so anything beyond that suggests a kernel issue.