How to continue the ZFS incremental storage replication (and avoid a full/base sync) after rolling back a guest

[H25E]

There is a way to continue PVE-ZFS storage replication as usually after rolling back a guest without having to do a full/base synchronization?

I haven't been able and I haven't find nothing on the internet neither. I have tried:

1. to prune dataset snapshots in the hot spare machine to match the main machine snapshots list
2. To manually create the @__replicate_10X-Y_ZZZZZZZZZZ__ expected snapshot in both machines, main and hotspare
3. To eliminate the synchronization job and create it again

First two points didn't work, returning errors (something about not finding the expected snapshot x, and "No common base to restore the job state")

Point 3 created a full/base sync from the beginning, that is what I was trying to avoid.

So, I don't know if there is a way to make pvesr continue doing incremental syncs from the rolled back point.

Thanks for your time.

 

[LnxBil]

Very good question. In general, your idea is correct and I use a similar approach in my own ZFS sync program outside of PVE, but PVE replication should be able to it its own. Maybe the staff can shed some light on this.

 

[H25E]

I think it's an interesting question too. Let's check if proxmox staff can provide some help here.

It would be helpul to know if there is a way to do it right now, and If not, if this is something that it's on proxmox roadmap or if it's clear that we are going to develop or own solutions here.