RC6 - Unwriteable Array

These are kind of strange:

 

May 26 23:00:23 TDS-Black shfs: /usr/sbin/zfs destroy -r 'cache-ssd/ARRs'
May 26 23:00:23 TDS-Black shfs: error: retval 1 attempting 'zfs destroy'
May 26 23:00:23 TDS-Black shfs: /usr/sbin/zfs destroy -r 'cache-ssd/carl'
May 26 23:00:23 TDS-Black shfs: error: retval 1 attempting 'zfs destroy'
May 26 23:00:23 TDS-Black shfs: /usr/sbin/zfs destroy -r 'cache-ssd/gaming'
May 26 23:00:23 TDS-Black shfs: error: retval 1 attempting 'zfs destroy'
...
May 26 23:19:41 TDS-Black shfs: /usr/sbin/zfs destroy -r 'cache-ssd/mtb'
May 26 23:19:41 TDS-Black shfs: error: retval 1 attempting 'zfs destroy'
May 26 23:20:33 TDS-Black shfs: /usr/sbin/zfs create 'cache-ssd/carl'
May 26 23:20:33 TDS-Black shfs: command failed: 1

 

Was it around this time the problems started, do you remember what you were doing at that time that would explain these? Any user scripts running?

This is the time I noticed the issue.  Every time it's happened, it's been when I was away from the system so no direct interaction from me would cause the issue.  I do have user scripts running.

 

Scripts

1. delete.ds_store - running hourly

2. Nextcloud OCC Preview - running daily

3. Nvidia Patch - running at array startup only.

 

I did migrate my SSD pool back to BTRFS last night and it still seems to be fine as of this morning.  That probably doesn't meaning anything though since it's only been a few hours since the last reboot.

Edited May 27, 2023 by csimmons

FYI, since moving back to BTRFS on both pools, the issue has not come back.

There's another report that should be the same issue, still investigating but possibly some interaction between the mover when there's other active that may cause the zfs datasets to unmount, do you remember if you were usually doing anything else manually or with a script at the time the mover ran? Your mover is set to run at 11pm.

I wasn't doing anything (manually) when it occurred. I did at the time have Unmanic running to encode some media as well a DS Store removal script that runs hourly.  Perhaps one of the two things could have complicated the issue.

If you can please re-test with rc7.

7 hours ago, JorgeB said:

If you can please re-test with rc7.

I'll try and give it a go this weekend.  Thank you!

Just to add that I'm not sure the issue is solved since I've not been able to reproduce, but possibly it happened with existing datasets created with an older zfs release, since you'd be creating new ones it will confirm or disprove that theory.

22 hours ago, JorgeB said:

Just to add that I'm not sure the issue is solved since I've not been able to reproduce, but possibly it happened with existing datasets created with an older zfs release, since you'd be creating new ones it will confirm or disprove that theory.

I'll keep you posted.