Re: [rdiff-backup-users] Safe to remove old increments while a backup runs?

[Dominic Raferd]

On 26/03/2014 09:41, Thomas Harold wrote:

On 3/26/2014 4:59 AM, Sam's Lists wrote:

I was removing some old increments on my backup server using:
 
rdiff-backup --remove-older-than 3M --force root

While this was running a cronjob on a remote computer started backing up
to my backup server (the same directory).

This is perfectly safe right?

Well, hopefully the remote computer would abort saying that the
destination rdiff-backup directory is in use... it's generally not safe
to have two rdiff-backup commands hitting the folder at the same time.
And I'm pretty sure that rdiff-backup checks for that.

In our backup scripts, we always do everything from the client end in
the backup script.  So the origin is in control of everything.  The
sequence of commands that the origin PC runs are:

- Run the rdiff-backup
- Run a verify on the last N increments
- Age out older increments

We also take a bit of care when rsync'ing the backups to the offsite
location.  We check whether there has been any activity in the
rdiff-backup directory in the last N minutes (usually 10) before we
snapshot the filesystem and send the rdiff-backup offsite.

Similar to your last tip, my offsite backup script aborts before taking the snapshot unless it gets null output from command:

ps h -C rdiff-backup|grep -E -v " (--restore|-r |--list|-l |--compare)"

The idea is to check that there are no ongoing rdiff-backup activities except for certain 'safe' ones such as restore, list, compare.

Dominic

--
TimeDicer: Free File Recovery from Whenever