Whats the best way to be able to backup mender server services atomically without causing any in-progress deployments from being cut off? is it even possible? I am using zfs and snapshots, but it seems to ensure data consistency mender services need to be stopped first before taking a snapshot?
As long as all the volumes are on the same zfs filesystem I don’t think you need to take any special care, snapshots should just work. Taking a snapshot is essentially the same thing as what would happen if the machine unexpectedly went down, and came back up again, and the DB is supposed to recover just fine from such incidents. That being said, I haven’t tested it myself, I’m basing this on theory.