zfs umount and spl_delay_taskq stuck for days #16647
Comments
arcenik
changed the title
|
It looks like the cause is that kernel thread arc_prune exited #16324 |
|
Can you try this patch pls? vpsfreecz@74964ac |
|
We have had issues like this recently on some of our systems as well, the call traces are similar. After that, any zfs or zpool commands hang indefinitely, until the system gets a hard reboot. |
|
@dennixxNL that looks like a different issue - can you try if this issue is fixed in newer releases? if it is it would be worth making a new issue for that fix to be backported to 2.1 |
|
I'd love to test, however until now I'm still unable to reproduce this issue in a test environment, and don't really fancy testing this in production. Also there are no newer RHEL8 packages available in the official repository except in the 'testing' repository, but that doesn't seem wise to me to use that in production. |
|
personally I wouldn't rely on any 3rd party packages and their assessment of what's stable, in my experience they tend to ship stuff with bugs that have been fixed... your updated stacks look like an issue I resolved a few days back though but I totally understand where you're coming from, I wouldn't want to risk production neither... how about a compromise then, could you try applying that one commit on top of what you're currently running? |
|
oh sh!t I mixed up what I'm replying to, actually the original issue from the OP here should be fixed by the patch, your stacks look differently and like something that's been fixed for a bit longer than this |
|
Is that fixed in 2.1 too? If so, could updated packages which include this fix be published for the RHEL repository please? |
Do you know if my specific issue has been resolved in the last week released version 2.1.16 for RHEL8 ? |
System information
Describe the problem you're observing
Zfs umount on an old snapshot (triggered by zfs destroy) command get stuck forever, until hard reboot.
Describe how to reproduce the problem
It just happens.
Include any warning/errors/backtraces from the system logs
First kernel thread [spl_delay_taskq] to get stuck:
Another [spl_delay_taskq] kernel thread:
An umount command umount -t zfs -n /data/archives/omega-2013/.zfs/snapshot/2013-11-27:
Another one umount -t zfs -n /data/archives/omega-2013/.zfs/snapshot/2013-11-28:
The text was updated successfully, but these errors were encountered: