Re: [rdiff-backup-users] Failed regression: missing snapshot

[Andrew Ferguson]

On Jul 10, 2008, at 9:31 PM, Dan McGuirk wrote:
> Also, I forgot to mention: it was about 8 hours between when the
> backup finished and when I (uncleanly) powered off the drive.  There
> were messages in the syslog from the time I turned it off about failed
> journal commits and so on.  I'm not sure why the journal wouldn't have
> been committed already after that amount of time.

Indeed. That sounds like the root of the issue. What filesystem are  
you using? Sounds like it might be worth reporting as a bug -- isn't  
recovering gracefully from power-failure the whole point of a  
journaling fs? :-/

> I wonder if the main problem was that the old 'current_mirror' file
> didn't show up as deleted.  I think the rest of the structure was
> probably intact.  Maybe I should have tried deleting that instead.

Yes, that is what the FAQ says to do to resolve this problem:

http://www.nongnu.org/rdiff-backup/FAQ.html#regress_failure

My apologies for not reinforcing that point before suggesting more  
drastic measures.



Andrew