Re: [rdiff-backup-users] Re: Restore stuck

[Blair Zajac]

Troels Arvin wrote:
> On Sun, 12 Feb 2006 15:47:21 +0100, Troels Arvin wrote:
>
> > On b, load avarage is high, and rdiff-backup has a resident set size of
> > 356MB (rising). On r, load avarage is practically 0.
>
>
> I tried using more potent hardware. It turned out that if I work on
> servers with more RAM, thrashing doesn't occur, so the restore-process is
> able to finish.
>
> Involved in the restore are directories with ~30000 files. I suppose that
> this is a hint about what's going on. Maybe rdiff-backup uses some
> somewhat "expensive" python container structures which makes rdiff-backup
> consume lots of RAM while restoring?

Also, directories don't scale too well with that many files in it.  On Linux, 
you can use the dir_index option when you build the filesystem.

Regards,
Blair

--
Blair Zajac, Ph.D.
<address@hidden>
Subversion training, consulting and support
http://www.orcaware.com/svn/