We're using rdiff-backup as part of a slightly
complex automated backup solution involving LVM snapshots and rsync.
Very infrequently, something goes wrong in
the preparation phase, like it did last night, and the result is that rdiff-backup
is run with a completely empty source folder, wiping out about 80 gigs
of data.
Now, the data is not lost or anything but,
as far as I know, if my backup runs okay tonight, I'm going to end up with
160 gigs of space used by rdiff-backup: the old history up to the point
where the folder got deleted, one history entry where all the folder contents
are gone, and then the start of a new history.
In other words, we seem to have only two
options: keep all the histories and doubling (or worse should it happen
again) our storage needs, or discarding the history up to the point of
failure.
Of course, we'll take precautions to make
sure it won't happen again, but either way, the question I'd like answered
that would solve everything:
Is it at all possible to undo the most recent
rdiff-backup run?