Restic prune takes many days with rclone + box

I have an approx 1 TB backup set on box.com. This is really a great feature of restic in many ways.

I notice that accessing by ‘restic mount’, ‘restic snapshots’ or ‘restic clean’ takes a modest amount of time but ‘restic prune’ takes many days (it was 3+ and counting but the tropical storm took out my connection so back to the beginning).

Anyway, is this expected?

I did see the GDrive question earlier but I think that something different is going on here.

Is there some tuning to make this faster for the rclone case. Any thoughts?

There are severe speed issues in current restic releases and some big improvements under development, see

Either try these proposed changes (with care!) or please be a bit patient.
The maintainers do a great job here and I agree that especially with pruning enhancements quality goes over development speed.

1 Like

Wow. Thanks for all your hard work and contributions!

I happy to wait; the backup itself checks fine, and I only wanted to clean up some unreferenced packs. It’s nice to see that this issue is being addressed.

Another approach you might consider:

Many of us use restic to backup to a local repository, which we then synchronize into cloud storage. (using rclone, ‘aws s3 sync’, etc.)

This provides several benefits:

  1. You have a local copy of your repo, for fast backups, restores and much faster prunes.
  2. Very efficient writes into the cloud.
  3. May allow you to move cloud data into high-latency/low-cost storage tiers (e.g. Glacier, GDA)
  4. Compliance with US_CERT 3-2-1 backup strategy

This approach has drawbacks as well, but is worthy of consideration.