Trouble with large backup

[Gavin Carr]

Hi David,

On Fri, Apr 24, 2009 at 05:30:34PM +0100, David Smalley wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> We've recently started to use Brackup to sync our bacula backup data over 
> to s3. After a couple of successful test runs we set it to work on our 
> 500gb~ of backups and it went on a few days.
> Unfortunately we had to kill the backup a few times and restart it while we 
> got traffic shaping in place to stop saturating our outbound connection. 
> Every night we had to kill the process and then restart it after the bacula 
> backups had completed.

While not ideal, brackup's been pretty resilient to that kind of treatment 
in my experience.

> Now when I come to try a test restore/list_backups/gc on the backup I'm 
> getting some funny results..
>
> $brackup-target -v amazon gc
>
> Unexpected line in metafile /tmp/1SQBmeinz5/o1m7MXxCX2, line 1: 
> ????n_???fsk???+?|?LiU??OK(y????Q??9??E??T??jv??w????|T0????1z?? at 
> /usr/local/share/perl/5.8.8/Brackup/Metafile.pm line 45, <$__ANONIO__> line 1.

Is the /tmp metafile encrypted? If so, unencrypt it first.


> $ brackup-target -v amazon list_backups
> Backup File                      Backup Date                      Size (B)
> - -----------                      -----------                      - 
> --------
> Argument "\x{32}\x{30}..." isn't numeric in numeric comparison (<=>) at 
> /usr/local/bin/brackup-target line 122.
> Argument "\x{32}\x{30}..." isn't numeric in numeric comparison (<=>) at 
> /usr/local/bin/brackup-target line 122.
> Argument "\x{32}\x{30}..." isn't numeric in localtime at 
> /usr/local/share/perl/5.8.8/Brackup/TargetBackupStatInfo.pm line 34.
> bacula-1240436350   Thu 01 Jan 1970 00:33:29          5388249
> Argument "\x{32}\x{30}..." isn't numeric in localtime at 
> /usr/local/share/perl/5.8.8/Brackup/TargetBackupStatInfo.pm line 34.
> bacula-1240512974   Thu 01 Jan 1970 00:33:29          5397846

This is fixed in the recently released 1.08. What version are you using?

Cheers,
Gavin