[rabbitmq-discuss] Hostname Issues with EC2 Cluster

[Kevin Kuhl]

Hi all,

I want to set up a 3 node cluster on EC2. We plan to set up a recovery 
configuration in amazon so that if we lose a server, we pull up a new one 
to replace it and cluster it. We;ve been able to set up a cluster manually 
by using the default hostname that resolves over DNS (ip-xx-xx-xx-xx); each 
of the machines can see eachother. This name however, will not work in 
production because it's the INTERNAL DNS reference which changes over time. 
If a failure occurs this will be different on launch and break the cluster 
(right?). 

The ideal situation would be to cluster these with static references to 
each machine. like rabbit at server1, 2, 3 where at any point a new one would 
come up and could join the cluster.

I read this 
article: http://karlgrz.com/rabbitmq-highly-available-queues-and-clustering-using-amazon-ec2/

Which works most of the way, but the part that's not explained is, once you 
set your own hostname and try to cluster them, how can they see eachother 
if the hostnames are not available over the EC2 network.

Hope this makes sense, any ideas or best practices are greatly appreciated!!

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130205/34ab153f/attachment.htm>