[wellylug] the eternal question - backups

[Richard Wade]

> On Fri, 27 May 2005, Richard Wade wrote:
> That's more-or-less what I currently do, but it doesn't provide enough
> 'history' - if I corrupt or delete a file and don't notice for a day or
> two, it finds itself on the backup.  That's why I am considering
> differential or incremental schemes.

I think I recall you saying something about wanting a push rather than a
pull (i.e push TO the file server), so how about this.. A script that runs
on the client that does the rsync to the server every night. A script that
runs on the server, timed to occur after the client has run, and that will
not execute unless a "lock file" has been "touched" in the root backup
directory for today's date. Then, have the script on the server do a
compressed tar of the directory to a tarball containing the date extraced
from the lock file? You could have this script just keep three (or however
many) tarballs before overwriting the oldest one. You could also easily
back up one or more of these tarballs to a CD-R, CD-RW or DVD-R from the
headless server.
--
rik