ZFS replication corrupts entire ZFS volume (Warning!)

[panz]

The system worked flawlessly with 9.2.1.3. I inserted the backup pool disks (n. 6 3TB WD Red) every friday, unlocked them and flagged "Enabled" in the ZFS Replication task.

The Replication took place overnight and I was happy :)

Then, I didn't replicate for a month and all started going bad

As suggested, I'm posting the required info in that issue.

 

[Michael Wyraz]

Is it right, that the destination volume (not the source) gets corrupted? If so, please change the topic accordingly because IMO it's a _big_ difference if I loose a replication source or target!

I would recommend the following test to find out if the issue is caused by zfs_send (which creates a corrupted stream) or zfs_receive (which destroys things during applying the stream):
- Use a different OS / ZFS implementation as destination. Zfs on Linux would be an option but this does not implement ACL, so the issue might not occur for that reason. OpenSolaris or Ilumos would probably a better choice here.
- Use a different OS / ZFS implementation as source and replicate to a freenas installation
- Try replicating from/to plain freebsd to determine if it's a freenas or freebsd-zfs problem.

Kind regards,
Michael.

 

[cyberjock]

This thread is old and dead months ago. Locking the thread.