georgeharley (3) [Avatar] Offline
#1
Probably an obvious observation but just in case anyone else falls into this hole...

In section 8.3.2 (Defining a policy) I found that when checking the Federation Status on cluster-b's management console the state column always just hung at "starting" and never said "running". Tailing the log file on the cluster-b host revealed a number of warning messages...

=WARNING REPORT==== 17-Sep-2015::16:39:17 ===
Federation exchange 'test' in vhost '/' did not connect to exchange 'test' in vhost '/' on amqp://<your cluster-a IP address here>
{error,etimedout}


To get around this you need to go back to the Amazon EC2 management console and add a new inbound rule to the cluster-a EC2 instance that allows the cluster-b EC2 instance to connect over TCP. I just added an "All TCP" type rule where the source was restricted to the IP address of the cluster-b EC2 instance. I also added a reciprocal rule for the cluster-b EC2 instance to allow all TCP from the cluster-a EC2 instance as well but maybe that is not strictly necessary. Once the rules were applied the warning messages in the logs stopped and on the cluster-b management console the Federation Status page's state value finally moved from "starting" to "running".

Hope this helps.