> I'm not quite following all of your statements. Specifically, I don't > see why you think you need a base AND a first snapshot. One snapshot is > sufficient for later restoring. Also, I don't think you need to > understand the underlying architecture in order to successfully use > snapshots (it might be helpful or interesting, but not really > necessary). The command 'snapshot-switch' reliably restores the state of > the CT it had at the time when you created the snapshot. It's as simple > as that. Also, when deleting snapshots, vzctl auto-magically does the > right thing in that it merges or deletes when it is appropriate without > affecting the other snapshots of the given container. >
Sorry about kicking this off again, but this hit to me over the weekend and it cleared things up for me. Please, someone, correct me if I'm wrong. As I understand it now, no single file corresponds to a snapshot and to think of it that way will lead you to do something silly with your data. A snapshot is an event. If you want to think of it in terms of files, it's the gap between the root.hdd and the delta. It's a like a HUP in I/O gives you a spot to go back to. When you delete a snapshot, you're deleting the event and this means the files that are either side of that discontinuity will be merged/healed. When you mount a snapshot, for example when doing file based backup ( https://openvz.org/Ploop/Backup) you're not actually mounting a snapshot because there is no snapshot file to mount; you're mounting the pre-snapshot data. Simon
_______________________________________________ Users mailing list Users@openvz.org https://lists.openvz.org/mailman/listinfo/users
