Posted by: Anonymous
on November 04, 2008 07:23 PM
Personally, I use rsync --link-dest=DIR (check out the manpage) to make a new snapshot, then run a script to run sha1 checksums on any file in the new snapshot with one link to it, and store the file in a pool under the name of the checksum, similar to how BackupPC stores its backups. If I was smart enough to use librsync for that one, I would; however, this seems to work acceptably well and doesn't take a great deal more time.
I've got that setup at work, with the snapshots available to a network of Macs via Netatalk running on a Debian Etch machine. So far, this setup has meant I can save several days' snapshots, are readable from Finder, and only unique files are stored more than once. I've even started to store workstation backups using the same mechanism (after building Darwin rsync on Etch, of course.)
I've tried TimeVault, but heck, to me it seemed like a less featureful version of BackupPC, with a GUI. No, thanks.