rabbitmq_clustering.txt

RabbitMQ clustering notes (Source: http://www.rabbitmq.com/clustering.html)

Before following these steps, install the single-node Xqueue first.

0) Stop all rabbitmq servers (at each instance)
    rabbitmqctl stop

1) Instances must be able to refer to each other by short names, e.g.

Instance1:
/etc/hostname:
    node1 
/etc/hosts:
    10.96.127.153 node1 
    10.96.127.154 node2 

Instance2:
/etc/hostname:
    node2 
/etc/hosts:
    10.96.127.153 node1 
    10.96.127.154 node2 

Source the new hostnames (if changed)
    /etc/init.d/hostname restart
Should also be able to ping each other

2) Instances must share same string cookie (copy one to the other)
    /var/lib/rabbitmq/.erlang.cookie

3) Restart the rabbitmq servers (at each instance)
    /etc/init.d/rabbitmq-server restart

4) Start the rabbitmq app on the original cluster node (e.g. instance 1)
   Note that it is started by default after rabbitmq-server restart
    rabbitmqctl start_app

   The nodes need access to port 4369 for empd (erlang port mapper daemon)

5) Join the cluster at instance 1 from other instances
    rabbitmqctl stop_app
    rabbitmqctl cluster rabbit@node1
    rabbitmqctl start_app

   The rabbitmq-server on node1 uses a port that's distinct from empd's 4369. 
   Not sure what the logic is for port assignment, but the empd on node1 will report what 
    port is used by rabbitmq-server

   To find the port, run fron node2:
    rabbitmqctl -n rabbit@node1 status