[rabbitmq-discuss] Federation configuration for clusters - failover

Wed Jul 25 12:41:08 BST 2012

Hi,

On 23/07/12 15:25, Laing, Michael P. wrote:
> I currently have the same configuration for each node in the cluster. Is
> this correct?

Yes.

> Also, what is the expected behavior when a node in the cluster goes
> down? If it is the node on which the federation plugin has created its
> queues etc, I would think it should recreate them on another node and
> keep on going. We will test this but it would be nice to know what to

Yes, the federation link will be re-established from another node in the
cluster. You can determine the active node by inspecting the 'status'
field in the output of

rabbitmqctl eval 'rabbit_federation_status:status().'

> expect. Perhaps we should we set the HA option? Currently losing
> messages is not an issue for us but en futuro it will be as we implement
> high volume applications.

If you are concerned about losing messages after they get routed to
queues then you should use HA.

-Emile