Dataset growing after restart of a failed replication!

[r00tb33r]

Curses!

A replication failed because dataset was busy...

I deleted the snapshot from the shell and started the replication again. Then the dataset in the pool started growing beyond the size I'm replicating!

I'm only using replication to migrate data from one TrueNAS server to another. How can the destination dataset grow beyond the size of the original!!!????
The original dataset is 8.61TiB, we're at 8.70TiB and still growing! Where did it find the extra data? It was going "x of 8.61TiB", but then it started going "8.62TiB of 8.62TiB" and so on.

I don't understand WTF it's doing and it will take days at a time if this is a trial-and-error process. I'd be done by now if I just did an SMB copy of my files...

 

[winnielinnie]

Are you using the same compression properties on both datasets, source and destination?

Do you have any residual snapshots / partial-transfers still on the destination pool? Sometimes you'll notice a strangely named dataset, like "mydata-34252".

Check with:
zfs list -t filesystem -r poolname

and

zfs list -t snap -r poolname

 

[r00tb33r]

No, it was a fresh destination pool. Yes, same LZ4 compression.

I ended up nuking everything and started over with a fresh pool again

Very disappointed in the replication integrity so far, I was expecting 1:1, not understanding why it went back to finished datasets on replication resume and grew them larger than they were originally.

I'll know how it went tomorrow night.