openstack team mailing list archive

Thread
Date

Re: Howto Nova setup with HA?

To: i3D.net - Tristan van Bokkem <tristanvanbokkem@xxxxxx>
From: Major Hayden <major.hayden@xxxxxxxxxxxxx>
Date: Tue, 14 Feb 2012 13:34:27 +0000
Accept-language: en-US
Cc: "<openstack@xxxxxxxxxxxxxxxxxxx>" <openstack@xxxxxxxxxxxxxxxxxxx>
In-reply-to: <20120214080751.7bcf812e@mail.i3d.nl>
Thread-index: AQHM6u/h3qtynZlorUCE7mTpvTIP0ZY8yXIA
Thread-topic: [Openstack] Howto Nova setup with HA?

On Feb 14, 2012, at 2:07 AM, i3D.net - Tristan van Bokkem wrote:

> So, we can run MySQL in master-master mode on multiple hosts, we can run nova-api on serveral hosts and load balance those and RabbitMQ has a cluster ha setup as well but is this the way to go? I can't find a clear answer to this. I am hoping one can shine some light on this!

If you want to make things a little simpler, an active/passive MySQL setup with pacemaker, corosync, and DRBD will work quite well.  ClusterLabs has a pretty decent howto on their site:

  http://www.clusterlabs.org/wiki/DRBD_MySQL_HowTo

As for RabbitMQ clustering, I worked with it for several days and had quite a few problems with data consistency and availability.  We eventually settled on the "legacy" H/A method for RabbitMQ using pacemaker and DRBD.  There's a pretty good walkthrough on RabbitMQ's site:

  http://www.rabbitmq.com/pacemaker.html

-- 
Major Hayden

References

Howto Nova setup with HA?
From: i3D.net - Tristan van Bokkem, 2012-02-14