Open fallenguru opened 4 years ago
btrbk needs a timestamp, which is set on the subvolumes created. Supporting configurable naming schemes would be pretty hard to implement, and would probably need a major redisign.
For example, I already use Timeshift [...] That means I already have scheduled (rw) snapshots of the root subvolume, I just need btrbk on the backup server to [re-shapshot them ro] to pull them
If the subvolumes to be backed up are read-write, it should be possible to create a btrbk config for this task, using wildcards. Something like:
volume /path/to/timeshift_rw_subvolumes
subvolume *
snapshot_preserve_min latest
snapshot_preserve no
target ssh://your-backup-site
IME btrbk does not distinguish between "its" snapshots and "foreign" ones as long as they follow the naming scheme. For instance, I use incremental strict
on my config to avoid accidental out-of-space problems. This requires me to do a manual btrfs send | btrfs receive
every now and then, and btrbk never complains or misbehaves about the results.
As I understand it, currently btrbk is designed to only work on snapshots it has created itself, or at least ones that follow the configured -- or is it any supported? -- naming scheme. This leads to duplicated functionality (= added complexity = more ways to fail) when snapshots created by other tools and/or manually created ones are added to the mix.
For example, I already use Timeshift for system-level undo on Ubuntu and Linux Mint -- it's well integrated and works well. That means I already have scheduled (rw) snapshots of the root subvolume, I just need btrbk on the backup server to [re-shapshot them ro] to pull them. Then I have a couple of subvolumes where before-change/after-change makes more sense than a time-based schedule, so I ro-snapshot them manually or via custom scripts, those just need intelligent replication. For the home subvolume, I'd appreciate the full functionality of btrbk, scheduled snapshots, backups, pruning, the works.