The distinction here between backups and archives becomes interesting, I think. We're using dirvish mostly for archives, with the expectation that in the event of a major (whole machine) outage, we rebuild the box and then import all our data.
As Andrew says, this is easier if you keep your working data separate from your binaries / system data. (It also helps if you have something like FAI and a clearly defined base build or SOE.) This can get complex, though, so you certainly do want to do some trial restores. On Debian you can get stung, f.e. if you underestimate the import of the /var/lib/dpkg hierarchy. If you're not running a 24x7 site I'd suggest you look at running your servers inside VM's (pick your poison) with perhaps weekly snapshots and daily/hourly dirvish runs. This is what I'm looking to do with our office file server. This brings me back neatly to the importance of distinguishing between backups and archives. Jedd. _______________________________________________ Dirvish mailing list [email protected] http://www.dirvish.org/mailman/listinfo/dirvish
