Closed araemo closed 4 months ago
If you are space constrained you can use zfs bookmarks, in this case you can still replicate from your source pool even if the snapshot on the source is destroyed (but a corresponding zfs bookmark exists).
Regarding your other points:
I am using sanoid and syncoid to do a daily snapshot of a space-constrained all-NVME pool, and replicate the snapshots to a bulk storage pool. It works well, but I'd like to see a bit of integration between the two.
Currently, I keep 2 days of snapshots on the source pool, and replicate (using existing snapshots only, no sync snap) the snapshots to the bulk pool.
If for some reason, the sync fails, there's nothing stopping sanoid from deleting the older snapshot, and that could get me into a state where I no longer have any snapshots that match on both sides.
I could probably 'fix' this by using a sync snapshot, but I am very space constrained on the source, so I'd rather avoid this if possible.
What I'd like to see is one of two things: