Zfs replication sends more data to remote than what is shown on source

[hescominsoon]

I am curous how this works. I have a freenas 11.3U4.x mahcine that i have replicating to a Truenas 12RC1 mahcine. right now according to the webUI and zfs get all total disk usage including snapshots is 2.44TB. However as I type this the target system shows 2.6TB and rising..and the replication is 87% completed menaing there's a significant amount of stuff left to send. Where is this other stuff coming from? Or is this a bug with the recipient station?

 

[develissimo]

Bump

We had the same issue since 11.x now we upgraded after some advice of the community to 12.0 TrueNAS due to the fact that lots of improvements had been done especially to replication... But no gain so far: The issue seems to be the same. The datatransfer process never stops until the receiving disk is full...


The output shows like:

So i guess sometime ago on 11.x there happend a replication issue on the receiving side and this issue seems not autocorrected.
We have a second receiving server for exact the same replication and on the second server replication works fine.

Only difference: The server on the receiving side where disk is filled up until out of space is on a far slower connection with only 20MBit ...
pool01 replication succeeds
pool02 replication succeeds
pool03 replication pushes data until disk full

So i guess we have to delete the whole replicated data on the receiving side and initiate a full replication beginning at zero?
Is there another solution?
Replication worked fine until ~ October.2020 ... would be nice not to have to transfer data which has already been successfully transfered.

Kind regards,
Raphael