[rabbitmq-discuss] Federation and Clustering...

[Simon MacMullen]

On 27/11/12 23:26, Steven Brooks wrote:
> On Tue, Nov 27, 2012 at 1:03 PM, David Buckingham
> <David.Buckingham at cbeyond.net <mailto:David.Buckingham at cbeyond.net>> wrote:
>
>     I’m not quite sure where I may have gone wrong.  By adding the new
>     clustered node as an upstream on rabbit1, I guess that I expected
>     the same configuration would be pushed-to/created-on rabbit3 that
>     was previously created on rabbit2.
>
>
>  From the documentation on Highly Available Queues
> (http://www.rabbitmq.com/ha.html) regarding using a cluster:
>
> "some important caveats apply: whilst exchanges and bindings survive the
> loss of individual nodes, queues and their messages do not. This is
> because a queue and its contents reside on exactly one node, thus the
> loss of a node will render its queues unavailable."

So putting it all back together, the OP needs to make sure that the 
queues created by rabbit1 on rabbit2/3 are mirrored.

If rabbit2/3 are running 2.x then you can do this by telling rabbit1 to 
declare them with x-ha-policy set to "all", using the ha_policy 
connection parameter.

If rabbit2/3 are running 3.0 then you just need to declare a policy on 
rabbit2/3 which matches the upstream queues and makes them mirrored.

Cheers, Simon

-- 
Simon MacMullen
RabbitMQ, VMware