Transferring with zfs send/recv--receiving pool is already 2TiB above sending and has not stopped

[okimdone]

So I'm moving a dataset to a new pool and the receiving dataset is 2TiB above the size of the sending dataset and still going. I'm at a loss whether I should stop the transfer and retry or just let it go to see if it stops.

Version: TrueNAS-12.0-U8.1

Commands used:

Code:

zfs snap Vol20TBx2/Media@migrate
zfs send -R Vol20TBx2/Media@migrate | zfs recv -F Vol14TBx8z2/Media

 

[okimdone]

So I restarted it and changed the command so I could track progress and I'm noticing something odd.

Code:

zfs send -R Vol20TBx2/Media@migrate | pv | zfs recv -F Vol14TBx8z2/Media

What pv is showing as the total data transferred so far is half of what TrueNAS is showing for the pools/datasets data usage...that's weird...I'm going to cancel this too and just transfer with rsync as I either discovered a bug or I missing something on how ZFS works...

 

[okimdone]

Also want to mention that after I killed the first transfer and it removed the dataset, that the disk still had all that data marked as used. While the used data was slowly going down I just decided to rebuild the pool

 

[sretalla]

I think much of the "problem" you have here is snapshots... you'll be sending files (or blocks that used to be part of files) that aren't counted in the size of the current dataset, but are in snapshots referenced in your recursive snapshot transfer.

Bytes over the network and bytes written to disk both apply different means of compression, so I'm not sure there's any fair comparison to do there.