Re: [rdiff-backup-users] Error in

[Mario Doering]

> >
> > rdiff-backup is started locally.
> 
> Not sure if we're missing each other. When I said log into the remote
> machine and run rdiff-backup locally, I meant locally on the other
> machine.
> 
> ie:
> 
> 1) ssh $REMOTEHOST
> 
> 2) rdiff-backup -v 5 --check-destination-dir $DESTINATION_DIR
> 
> rdiff-backup is now running against the same files as in your original
> post, but it is now able to set extended attributes, create hardlinks
> etc, because it has direct access to the ext3 filesystem, rather than
> going through the limited virtual filesystem under the cifs
> mountpoint.

You mean locally on the computer which contains the backup-up files
(including riff-backup-data) shared via samba?

If yes, a login there is not possible. It is a terastation (=black
box) with only access via cifs.

And this share is mounted on the linux system running rdiff-backup
(which I meant with locally, cause the data it wants to back up is on a
remote host :) )


> >I cannot install nfs on that machine,
> > but I can rsync all the stuff to another directory (ext3) and run
> > the check command there.
> >
> 
> That should work also. But it won't fix the warnings (inability to do
> xattrs etc) you'll still be getting by trying to run rdiff-backup onto
> a cifs mount.
> 
> I have another question: Was your rdiff-backup to a cifs mount working
> before? It's possible that rdiff-backup works fine to a cifs (with
> warnings about fs limitations), but the regressing doesn't. There's a
> fair chance that the regressing logic isn't as well tested as the main
> backup logic. In which case you will need to remember to ssh into the
> remote server to fix rdiff-backup problems.


Ah yes.. I should have give more information on that.
The backups themselves work great. I guess that a backup session got
canceled in between .. maybe because of a network error or a reboot.

So it seems to me as well, that only the regression code is missing some
of the backup-cifs-features :)


@rdiff-backup crew:
Is this info enough for the coders to know where the problem is located?


> My suggestion is that you change your setup into a "pull" rather than
> a "push". ie install rdiff-backup on the remote host (backup server I
> presume) and run it there instead.


It is set up that way, as written above :)