Just ran out of space so ran a prune which deleted a whole bunch of snapshots but has now failed because there is no room to fossilize.
What’s the best way out of this catch 22? I’m happy to completely delete the data in the deleted snapshots. Can I just add a -delete-only flag?
I would add more space, not delete existing data!
Or you can run prune in exclusive move. It assumes, as the name implies, exclusive access to the storage, and bypasses two-step fossil collection. prune · gilbertchen/duplicacy Wiki · GitHub
Unfortunately adding space isn’t an option.
Is there any way to tell how much free space prune needs to run?
If even file renames don’t work, nothing will help. Run it in exclusive mode.
Okay, an exclusive run seems to have given 3 GB space.
What’s best practice for my future runs? Run as normal without exclusive? Do I have to do a backup in between or not necessary?
What are the downsides to always running in exclusive?
Data loss, unless you always 100% guarantee that no other duplicacy instance is touching the storage while exclusive prune is running. Exclusive mode turns off all safety and concurrency promises. Use it only when there is no other way to get of the pickle.
Instead setup notifications when you approach quota. Better yet, increase quota, or move your backup elsewhere. Duplicacy backups are incremental, if you are pushing the limit, prune can only buy you so much time in exchange for losing versions. It’s a very bad trade.
From someone who’s run out of space a few times…
Make sure Duplicacy didn’t upload any 0-byte chunks. Manually delete them if it did then do another check.
From now on, prune to a tighter retention to make sure you never run out. -exclusive won’t help you here and is dangerous, but doing a once-only -exclusive -exhaustive will clean up unreferenced stuff. Enable -zstd with best compression may buy you more time, but ultimately it sounds like you need more space!