Question about data on target after recreating new snapshot tasks

[vafk]

I would like to clarify perhaps a basic question which is not yet clear to me.

I run a base Freenas with two "backup" Freenas. After a recent upgrade to 11.3 I deleted all snapshot tasks on source and would like to set up new tasks. The data replicated to backup1 and backup2 is almost up to date.

Question: After I will create a new snapshot task on source and send over to backup1 and backup2 will it preserve the data already existing on target and add/update/delete data or will the complete data from source be copied to target regardless of the already existing data on target?

One of the reasons I am asking is that the size of one example dataset on source is 360 GiB while the size on destination is already 460 GiB. There are no snapshots on target and the first snapshot that is currently running is constantly growing the size of target and I have no clue where the difference of 100 GiB comes from...

 

[Basil Hendroff]

Question: After I will create a new snapshot task on source and send over to backup1 and backup2 will it preserve the data already existing on target and add/update/delete data or will the complete data from source be copied to target regardless of the already existing data on target?

The latter is true. Btw, we have very similar configurations.

 

[vafk]

@Basil Hendroff - thank you!

So what should I do to keep my data on target2 and only update the difference? It feels hard to accept that all the existing data (ca. 7 TiB) on target needs to be deleted and sent over again just because I reorganized my snapshot tasks on source...

*** Edit

I think I found the explanation here:

Reddit - Dive into anything

www.reddit.com

 

[Basil Hendroff]

Not delete the snapshot tasks!

Extract from the User Guide Replication section :

The replication task asks to destroy destination dataset snapshots when those snapshots are not related to the replication snapshots. Verify that the snapshots in the destination dataset are unneeded or are backed up in a different location! Allowing the replication task to continue destroys the current snapshots in the destination dataset and replicates a full copy of the source snapshots.

I've kept legacy replication tasks for this reason and haven't explored new replication features as yet. SSH+NETCAT transport holds the promise of high-speed replication. I'd be interested in your experiences with the new replication engine.