Hi,
In the case of ZFS, when using the upcoming new feature of “refresh” copy which is incremental will the container be snapshotted and the snapshot be sent via zfs send?
Cheers,
Jon.
@monstermunchkin would be able to confirm the exact logic used, but yes, the plan is to use our different backend drivers similar to normal copy/move to sync the container and snapshots.
For ZFS (and most other backends), this would likely involve first syncing all snapshots, then making a new temporary snapshot, transfer that and restore it on target.
ok, just wondering as one of the benefits of ZFS is you can snap shot it while running and just send the diff via zfs send to another host which makes it useful for incremental updates (assuming you already did a full send initially). At the moment I’m doing it with sanoid/syncoid which is clunky, so would be cool to be able to do this with built in lxd commands.
Cheers,
Jon.
In Lxc copy --refresh workaround: efficient incremental ZFS snapshot sync with send/receive I posted a script for the regular and efficient sync of containers to hot/cold standby clones based on ZFS storage volumes and zfs send/receive . This is working very well in test since 2 months, I will use this in production soon.