On Thu, May 31, 2012 at 4:43 PM, Jeff Gipson <jeffagipson@xxxxxxxxx> wrote: >> Where do you place your backups, Jeff? :-) > > Well, I just started using SpiderOak. Jury's still out. Their > deduplication and compression ratio seems pretty good (185GiB on my disk > becomes 145GiB in their system), and they support file versioning, but > don't seem to have a point-in-time restore feature. I think I can create > a script to do that, but I liked my old rsync backups better. > > Also, you either have to run their GUI client *all* the time (or you > guessed it, no scheduled backups run) OR you have to run the client via > Cron in a headless sort of mode, when means (you guessed it... no GUI). > > You can't run multiple instances of the client. > > This limitation hasn't been an issue for me. I ran the GUI to configure > my backup selections, then quit the GUI and use cron to run the backups > on my schedule; But I can see this being a show-stopper for some people. > > If I hadn't just bought a year in advance, I'd probably run the old > rsync script that I used to use which send the data over a VPN (or ssh) > link to a remote system. It created a hardlink structure which was handy > for point-in-time restores, i.e. on the remote system: > > backups/2012-01-01/DriveC > backups/2012-01-01/DriveD > backups/2012-01-02/DriveC > backups/2012-01-02/DriveD > > etc... > > With the exception of the first backup, all backups were incremental. > The result was that you could enter the folder for whatever day you > wanted and see all the files as they were on that date, not just the > ones that changed and were backed up. The rsync command was a monster: > > /usr/local/bin/rsync -az --stats --timeout=300 --progress \ > --delete --delete-excluded --exclude-from="$6" \ > --log-format="%t %f (%l/%b)" \ > --link-dest="$DESTRT$PREFIX_TAG-Previous" "$RSYNCSHAR" > # then > ln -sf "$DEST" "$PREFIX_TAG-Previous" > > Of course the script set many of the other variables. I'd probably write > this script slightly different now (this script was written 8 years ago) > but the principle is the same. > > 1) Do an initial backup with rsync an set the "previous" symlink to > point to the destination folder > 2) Next day, create a new date-code-based destination folder, backup > using rsync against the previous destination folder (see --link-dest) > 3) Update the "previous" symlink to point to the most recent backup > 4) repeat steps 2-3 > > My script was also kinda ugly, but it ran for years, and with the links > neatly arranges as described above, I was able to configure an apache > server to allow other users to browse and download old versions on their > own. > > It's not perfect, I'm sure, but here's the whole script in context (for > educational purposes) > > http://fpaste.org/ry7G/ <- note This paste will self-destruct by > tomorrow. Thanks, Jeff, for your so detailed and helpful answer! Paul -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org