[rabbitmq-discuss] RabbitMQ clustering and node failure recovery strategies
matthew at rabbitmq.com
Wed Oct 6 12:01:58 BST 2010
In general, please don't think of clustering as trying to provide HA.
Clustering is solely about scalability.
For HA, the only easy thing that currently exists is active/passive
which is documented at http://rabbitmq.com/pacemaker.html
In that guide, it does go so far as to explain how you can have a
cluster of nodes, in which each node is actually an active/passive HA
pair, and thus achieve both the scalability of clustering, with HA.
More information about the rabbitmq-discuss