Re: [rdiff-backup-users] external hard disk backups - best practice

[Jakob Unterwurzacher]

Chris Wilson schrieb:
> What I've started doing to work around this is to rsync the original 
> data (not the rdiff-backup copy) offsite and run rdiff-backup locally on 
> the remote server. Not perfect, I end up with two copies of all the data 
> on the remote server, but at least it does work and appears to save 
> bandwidth.
>
> Cheers, Chris.

Another interesting approach.

> On Mon, 17 Mar 2008, Greg Freemyer wrote:
>> The bad news is that if my first set gets corrupted, then rsync will
>> relay the corruption out to the offsite copy.  My reason for two
>> copies is disaster recovery, not backup repository corruption.  May
>> need to rethink that based on this discussion.

I approached the corrupted backup problem by keeping multiple versions 
of the first backup (the rdiff-backup copy) on the remote server, with 
identical files hard-linked as described here: 
http://www.mikerubel.org/computers/rsync_snapshots/ (basically, it's the 
--link-dest rsync option).
All the work is done by rsync, so this is as bandwidth-efficient as it gets.
For me, this takes only little more space than keeping a single copy, 
but that depends on the volume of changed files of course.

Jakob