[rabbitmq-discuss] Message Priority support and HA Failover

[Matthew Sackman]

On Mon, Jul 12, 2010 at 12:01:17PM +0100, Simon MacMullen wrote:
> On 09/07/10 18:51, Timothy Chen wrote:
> >I've been reading more about the clustering and ha failover of rabbit
> >mq, and sounds like there is no out of the box solution for live
> >failover when a node fails? I see that the routing information will be
> >persisted and can be transferred over to another node, but messages will
> >not?
> 
> Unfortunately that is the case. We're aware that clustering / HA is
> a weak area at the moment.

The active/passive setup *will* recover persisted messages that have
been written to disk. http://www.rabbitmq.com/pacemaker.html has the
full details. There is no active/active support out of the box, but a
number of possibilities are available here with support from clients,
such as publishing everything to multiple brokers with identical
topologies.

Matthew