Feature Request: extra schedule for prune

[tgruenert]

at a 15 TB archive a 'prune' takes a very long time. this run´s almost after every backup.

it would nice to have a configurable schedule for prune independent from the schedule for backup. In my case a prune is not necessary after every backup.

[micw]

At the moment, due to single scheduling, we cannot get any conflicts between prune and backup. If we have multiple schedules,

• should we ensure that prune/backup do not run at the same time?
• should we prune only if backup succeeds (to avoid that if a backup fails, all other backups are pruned over time)?

Another way (with potentially fewer side effects) would be to have some option that checks how many recent backups are there and prunes only if that exceeds a certain number.

Edit: is this only about the "prune" command which is part of the "forget" run or is it for both, forget and "prune"?

[tgruenert]

Thank you for your questions. I would follow the idea of prune / forget after x successful backups.

I think scheduling should be done for prune because this is the time consuming operation. Forget can run every time after backup.

[micw]

The current master contains a version with separated prune. It is usually ran after the beackup, except you pass a crontab like `--prune '0 19 *'``

Note that a backup is skipped if a prune runs on it's schedule and vice versa. There is an example in the docs.

Also a new option RESTIC_PRUNE_TIMEOUT has been added that allows to terminate pruning if it takes too long.