On Fri, 2012-03-02 at 09:08 +1300, Adrian Mageanu wrote:
> On Fri, 2012-03-02 at 09:02 +1300, C. Falconer wrote:
> > Adrian Mageanu wrote, On 03/02/2012 08:57 AM:
> > > On Thu, 2012-03-01 at 21:26 +1300, Chris Hellyar wrote:
> > >> So..  Come on folks, what's cool?
> > > Cool in my opinion is this poor man's cvasi-"high availability" solution
> > > I'm using at a site: two Debian boxes, with several VMs each, mirrored
> > > via only a network cable connected straight from NIC to NIC.
> > >
> > > Disclaimer: I did not set it up, but I can ask the author if you want to 
> > > know more.
> > How do you make the disk highly available?   drdb?
> > 
> 
> And here is the author to explain this elegant solution:
> 
> Steve?...
OK, there's a couple of chunky debian boxes, running loads of memory
( well, with the price of it at the moment, it would be rude not to! )
and a hex core CPU. One live, one backup. Each will eventually be set up
the same, but it's also a time to upgrade from 10.04LTS to squeeze so
it's not at the moment.

I use the physical box as an infrastructure server, and then use KVM VMs
to publish different aspects to the office / outside world - web, mail,
samba, etc. Each of these VMs has an external IP address, but also a
private virtual subnet all of its own.

The backup server is set up with the same VMs, same external IP address,
but a different virtual network. The primary network interface of the
backup server is not connected.

The two private subnets are connected across a dedicated network
connection ( just a cable at the moment! ), and the backup server rsyncs
regularly with the live one using this network.

It's not a particularly elegant solution, but 

 - it had to be reverse engineered into the existing system, 
 - the switchover had to be simple to perform ( swap the cable
across! ), 
 - having to encrypt all of the data in a way such that theft rendered
it useless complicated stuff as well.

In addition, either system can function happily on it's own. It's not
perfect yet, but it is getting there.

Right beside me is part 3, which is an offsite backup that syncs up as
well. I ran out of data cap last month so it's had to wait... (:

I'd appreciate any comments ( apart from the not implementing a full
cluster ones. My choice! ) on how to improve this. It's for a local
charity doing extremely worthwhile work. 

Hope that makes sense,

Steve

-- 
Steve Holdoway BSc(Hons) MNZCS <[email protected]>
http://www.greengecko.co.nz
MSN: [email protected]
Skype: sholdowa

Attachment: smime.p7s
Description: S/MIME cryptographic signature

_______________________________________________
Linux-users mailing list
[email protected]
http://lists.canterbury.ac.nz/mailman/listinfo/linux-users

Reply via email to