[SOLVED] Dealing with corrupted chunks

B

burgekin

New Member
Nov 8, 2025
5
0
1
Hi there,
I recently had a funny kernel panic on my system and that seems to have introduced some some corruption in some chunks in my zfs pool. As a result now all backups for a group are failing verification as zfs just returns IO errors for those.

Is there a way to find out which backups are using which chunks? As far as I understand it, all newer backups depend on older backups for duplicate files, so a corrupted chunk in an early backup will corrupt all subsequent backup as well.

I'm currently running a zfs scrub, but if that won't be able to fix my issue, what's the recommended way to go? I'd prefer not having to delete all my backups...

Please don't ask me about my setup. I'm running an unofficial PBS docker container and I'm well aware that I won't be getting any support for issues with that. My questions should be unaffected by this.
 
Last edited:
_gabriel said:
one way is doing a new backup which will readd missing/corrupted chunks.
Click to expand...
Any PBS won't complain about missing chunks in that case? I ultimately went the full-delete and new backup route because I frankly felt very uneasy without any backups and surprisingly also got an error about corrupted chunks with my first verify job. I deleted one more corrupted chunk and now it seems to work fine but yeah.
 
What I feared was ultimately true, PBS does now indeed complain about missing chunks.

I did however find this reply so I hope that solves my issue, even with missing chunks now.
 
more and more backups verified as bad so I had to wipe my entire stash of backups, not just for one group. I recreated the datastore and ran into more errors trying to create new backups.

The kernel panic and by extent also the corruption were caused by a faulty memory stick. Unfortunately I don't have ecc.

With the faulty stick removed now everything is working.

Thanks so much for your help!
 
Last edited:
You must log in or register to reply here.
Top Bottom