Re: [rdiff-backup-users] recovering from no space left on device

[Steven Willoughby]

Michael Ross wrote:
> My recent backup failed with a "no space left on device" error 
> (rdiff-backup 1.2.7 on Linux). It was running as root, so the partition 
> (which contains nothing but the backup) is completely full. Therefore, I 
> can't run the suggested "--check-destination-dir" command because it 
> reports "[Errno 28] No space left on device: 
> '/mnt/backup/rdiff-backup/rdiff-backup-data/rdiff-backup.tmp.0'".
>
> I've looked through the mailing list, so I know this is a difficult 
> situation to recover from. Are there any recommendations of how to 
> successfully run "check-destination-dir" in these circumstances? Is 
> there anything in the rdiff-backup-data directory that is safe to delete 
> to make room?

I imagine it's safe to remove the file_statistics* files but have never 
tried it.

You could also look for large increment files and delete them on a 
case-by-case basis if you decide you don't need them.  Obviously in this 
case you won't be able to restore the files in question at or before the 
date of the increment you delete.

Finally, you could emulate --remove-older-than by removing all files 
with the timestamp of your oldest backup.

> There are other partitions on the machine that have free space (though 
> not enough to hold the entire backup).

You could temporarily move old increments to one of those partitions. 
Something like:

mv rdiff-backup-data/increments/$dir.2008* /tmp/babies

rdiff-backup --check-destination-dir .

mv /tmp/babies/* rdiff-backup-data/increments/

I've never tried this, either, but I can't imagine why 
--check-destination-dir would care.  :)

Steven