[rabbitmq-discuss] Portability of the RabbitMQ database

[Matthew Sackman]

On Mon, Jun 21, 2010 at 01:15:30PM +0200, Andreas Jung wrote:
> You can not use rabbit at localhost for a RabbitMQ cluster since you need
> distinct node names..how would you move a database from a cluster node
> to another server?

Again, this is covered in the guide at
http://www.rabbitmq.com/pacemaker.html#ha_cluster 

"Rabbit nodes within a cluster need to be able to resolve each others'
host-name. Currently we have that the node-name is set to
rabbit at localhost which must be changed as the other nodes in the cluster
would get the wrong idea if we asked them to resolve localhost. We can't
use the IP address raw (i.e. we can't have the node-name as
rabbit at 192.168.50.1) without switching to long-names, which is more
work, so instead our plan is to give a host-name of rabbit-ha-1 to
192.168.50.1, rabbit-ha-2 to 192.168.50.2 and then set the node-names to
rabbit at rabbit-ha-1 and rabbit at rabbit-ha-2. There are a number of ways to
do this - either configure this in your DNS server, or edit enough
/etc/hosts files to ensure that all the HA nodes can resolve rabbit-ha-1
and rabbit-ha-2."

Matthew