Re: [rdiff-backup-users] Staged backups

[chuck odonnell]

Ian Jones wrote:
> 
> Hello. I have about 20GB of data to backup from a remote site. 
> Clearly, it's not practical to do this over the Internet in one go, 
> so I would like to stage the backup over several sessions. So, my 
> question is: what it the best way to do it? If I backup, say /dir/a, 
> then subsequently /dir/a and /dir/b, will /dir/a get copied a second 
> time?
>
> An alternative approach would be to make a preliminary backup on DVDs 
> and copy the files to the backup machine. If I then use rdiff-backup 
> to do incremental backups, how to I ensure that the files that are 
> already there are not copied again, i.e. how to I add them to the 
> archive?
>

i typically use rsync for the initial copy, then once i have a
complete (albeit now out of date) archive, i run rdiff-backup manually
one time with the '-force' option to let it create its housekeeping
structures and sync up. after that you can script/cron rdiff-backup
normally.

if the concern is hogging bandwidth on your initial copy, either copy
one folder at a time at on your own schedule like you suggest above
(but using rsync, not rdiff-backup), or use rsync's '--bwlimit=n'
option to keep other data moving. i prefer the former, breaking it up
by folders, but for reasons other than bandwidth management.

i will typically create a shell script that calls rsync on one folder
after another and run it through 'screen' or 'at' so it backgrounds
nicely without having to leave a terminal open. the reason i like to
break up the archive into smaller chunks is so rsync doesn't have to
overwork itself too much juggling a zillion files in its initial
discovery stage, plus you have clean built-in checkpoints in case
something goes awry.

if you do break it up into chunks, don't worry if they are out of date
with each other. your goal is to get the initial bulk copy done. the
subsequent 'rdiff-backup -force' will sync everything up nicely.

best,

chuck