[Dirvish] singling out crucial data

Jason Boxman jasonb at edseek.com
Tue Jun 23 21:31:04 UTC 2009


----- "Loren M. Lang" <lorenl at alzatex.com> wrote:

> Jason Boxman wrote:
<snip> 
> > 
> > I don't want to backup the entire host hourly, as I do so nightly
> anyway.
> > Would I want to create an entirely new 'host' for handling this, or
> would it make
> > sense to finally investigate the branch feature of Dirvish, or some
> other way?
> 
> I create a separate vault per partition per server and, in some
> cases,
> multiple vaults per partition.  I use xdev and exclude to ensure that
> there is no overlap in backed up data.  I recommend putting data that
> is
> backed up more often into it's own vault and just running dirvish on
> that vault as appropriate.  I have never looked into branches and am
> not
> sure when they would be used.

I did that initially, but found it annoying to do a full metal restore as I needed to sync over each partition instead of a single vault tree.  (I don't have a build server, so I either restore from backup or reinstall.  I prefer the former.)  I suppose I could `cp -al` it to another local directory, then restore from that...

What naming scheme are you using?  I'm using foo.domain.net for my vaults.

Thanks!



More information about the Dirvish mailing list