[Gluster-infra] Backup strategy for our infrastructure?

Justin Clift justin at gluster.org
Mon Sep 1 04:32:48 UTC 2014


On 01/09/2014, at 4:43 AM, Louis Zuckerman wrote:
> Following up on my experiments in rackspace...
> 
> Scheduled server image is by far the easiest thing to use. No software is required on the machine and it's trivial to restore the whole system. The platform can make automatic daily images & prune old ones.
> 
> Backups are a bit more involved, requiring an agent installed on the machine. As far as I can tell files can only be restored to the original backed up machine, which is inconvenient. I would like the ability to restore files to a clone of the original machine, or somewhere else entirely. I've opened a support ticket asking about restoring to a clone & will update when I have more info on that. One thing I noticed about restored directories: they don't delete new files, only restoring the state of backed up files (recreating them if they were deleted). IMHO, there should be an option to delete any other files not in the backup when restoring, so the restored directory structure is identical to when it was backed up.
> 
> In summary, I strongly suggest that we immediately enable scheduled image creation for all of our cloud servers. That will take just minutes to set up and provides an essential safety net.

Cool.  Are you ok to set that up?

I took a look at the Rackspace agent enabled backup, but discounted
it as it explicitly only works for Rackspace VM's.  We still have
infrastructure in iWeb (Gerrit and Jenkins), so we need something
that's not Rackspace exclusive.  At least for now.

I've created a basic VM in Rackspace to hold backups
(backups.cloud.gluster.org) and have setup MediaWiki to automatically
backup it's database and files to that each night.

I'll get to the other services we have soon, one by one, and ensure
they're taken care of as well.  eg WordPress is likely next, then
I'll tackle Gerrit and Jenkins.  (those two scare me a bit ;>)

Everything else can come after that. :)

+ Justin

--
GlusterFS - http://www.gluster.org

An open source, distributed file system scaling to several
petabytes, and handling thousands of clients.

My personal twitter: twitter.com/realjustinclift



More information about the Gluster-infra mailing list