[Gluster-infra] Backup strategy for our infrastructure?

[Justin Clift]

----- Original Message -----
> Le mercredi 13 août 2014 à 09:24 -0400, Justin Clift a écrit :
> > Hi all,
> > 
> > I'm kind of worried that (as far as I know) we're not backing up
> > our GlusterFS infrastructure (websites, mailing list, various
> > applications, etc).
> > 
> > Does anyone know of stuff we do backup at the moment?
> > (just in case I've missed something)
> > 
> > If it turns out we really aren't backing anything up, what's a
> > good solution for us to look into?
> 
> From others projects I manage, I back everything on internal systems of
> RH. But I do not like the solution as this does prevent community from
> helping.
> 
> I would suggest do that on a rackspace server somewhere, but then we
> have a issue if someone get the access and erase everything from the
> interface, so I am not that thrilled by the idea.
> 
> What about:
> - make a server for backup on rackspace
> - make sure it is hardened ( since it will likely have root access
> everywhere to read all file )
> - set some backup ( rsync, anything )
> see with internal RH IT to make offline backup ?

Sounds workable to me.  This makes sure we've got a backup that's
online, and also one that's offline just in case. :)

Kinda wondering if we should consider a backup program like BareOS
too, though it could be overkill.


> First, we need to have a inventory of what we need to save, and for how
> long so we can decide the size of the server and offline backup.

These pieces spring to mind immediately:

 * Gluster Forge (repo data + all Gitorious config)

   Unless we're sure Gitorious AS is backing it up well enough.

 * Gluster Wiki

 * Jenkins (configuration, and all of the build artifacts)

   The user list + authentication for Jenkins is its hosts /etc/passwd
   file, so backing up the auth info for the local box will capture
   that ok.

 * Gerrit

 * Our Mailman mailing lists (config and data)

There may be other stuff, but those are the ones I can think of
right away.

For backup duration, I'd say we probably want to have dailies for
about 2 weeks, weeklies for about 2 months, and monthly ones after
that for at least 2 years.

+ 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