ZFS silent data corruption question

[pos-sudo]

Dear,

I just read about the silent data corruption bug in OpenZFS here: https://github.com/openzfs/zfs/issues/15526

I was wondering since this is an ancient bug in ZFS if we need to take immidiate action on Proxmox Nodes that have OpenZFS version below 2.2.0? Since I read on the internet that the bug is there for a long time but it only appears to be exploited since the latest 2.2.0 and 2.2.1 release. And that it theoritically could be exploited in lower versions but that in a real world scenario this only appears if you have specific workloads.

We have the following setup on most servers: RAID 1 ZFS Mirror LZ4 compression and disks in raw file for KVM VM. So my question is do we need to be afraid of this bug?

 

[leesteken]

You might want to read up on all the threads about this on the forum.

 

[pos-sudo]

You might want to read up on all the threads about this on the forum.

True I recently did but in my opinion I couldn't get a hard answer to this question above

 

[leesteken]

True I recently did but in my opinion I couldn't get a hard answer to this question above

Either you were on a version with this issue for a long time (and it did not happen) or just dist-update to version 2.2.2 (no-subscription) and you'll be fine also.

 

[esi_y]

So I also wonder, now getting zfs 2.2.3 and looking at this:

https://github.com/openzfs/zfs/issues/15933

Either you were on a version with this issue for a long time (and it did not happen)

What kind of logical reasoning is this for silent corruption? Then I might as well be on filesystem without checksums and only worry about hardware CRC errors (save for which, "it did not happen").