`borg list ./` hangs (or is really slow)

[zhou13]

I have a borg repository that is around 700G, previously managed by Pika Backup. Recently, I noticed that Pika no longer backup anything into this repository. I took a look with borg and it seems that borg list ./ or borg check ./ just hangs there for at least an hour. glances command shows that sda1 (where the backup resides) has IO about 90M's, which looks normal. Is there a way to debug the problem to save this repository?

[ThomasWaldmann]

Are you using a non-default lock wait time and the repo is locked?

The default is to only wait 2s (or so), but that can be changed via an option.

If a stale lock is the problem:

Make very sure nothing is accessing the repo and then use borg break-lock REPO and try again.

[ThomasWaldmann]

In case you meant borg is actively doing 90MB/s I/O on that disk, then it's not a locked repo.

It could be borg replaying segments due to a failed transaction, then you just have to wait until it is finished.

[zhou13]

@ThomasWaldmann Thank you for the response. It would be better borg can print something like borg is replaying segments, this might take a while so that the user has correct expectation.

After about 2 hours, borg list ./ works now! For the record, now if I execute borg list again, it shows results instantly.