Backup strategy for S3?

Martin Atkins mart at degeneration.co.uk
Thu Apr 30 23:45:30 UTC 2009


My strategy on Debian was to back up /etc, /home and the output of "dpkg 
--get-selections". The latter can be used to make dpkg install all of 
the software you had installed.

You might like to back up /var too. I didn't include it because the 
interesting parts of it were backed up by more specific backup targets 
on my servers.

I did this by writing the output of dpkg and some bind mounts into a 
temp dir and backing up that temp dir.

Unfortunately I no longer have the script I wrote to do this, but it was 
pretty straightforward.

Ed Blackman wrote:
> I'm exploring brackup as a replacement for my curent homebrew "monthly 
> full backups to DVD-R, incremental daily backups to CDRW" backup system.
> 
> I plan to use the Amazon S3 target because I want to add reliable 
> offsite storage, and am willing to pay for it, but at the same time 
> don't want to pay to store bits that are already stored offsite in a 
> reliable way: unmodified files that my Linux distribution provides. 
> (Ubuntu, in case it matters)
> In my current system, I do a lot of processing to remove unmodified 
> distro files (to save space, instead of money) from the list of files to 
> back up (and just dump the package and version of all distribution 
> packages I have installed).
> 
> I don't see an easy way to hook that into brackup, though, and wanted to 
> see what other people do to backup files in /usr, /var, /etc, and other 
> system directories.
> 
> Things I've considered include:
> - creating .brackup.conf from a template to programmatically add ignore= 
> lines for each unmodified file (in the 1000s of lines)
> - patching Brackup to add a "ignore-from-file" configuration directive 
> that would accomplish the same thing
> 
> Is there anything that I'm missing?  Any feedback would be appreciated.
> 
> Ed



More information about the brackup mailing list