[rabbitmq-discuss] Distributing clients across a cluster

[Matthew Sackman]

On Fri, Jan 14, 2011 at 07:57:29AM -0800, Bill Moseley wrote:
> On the same topic, I'm also curious how clients are suppose to behave.  If I
> have workers listening on a queue how should they handle a network failure?
>  Do any of the clients manage reconnecting on error?

Not out of the box, no, but then I think things like Spring-AMQP have
this functionality, or something close to it. You should be able to
fairly easily get notifications back from the client library when the
connection drops in all of our clients, so at that point, reconnect and
ensure any resources you need do exist.

> BTW -- I also looked at the HA docs at
> http://www.rabbitmq.com/pacemaker.html -- but that's replicating queue data,
> correct? Is there a need for DRDB if willing to accept loss of messages?

If you're simply looking for availability of service then a Rabbit
cluster is likely your best bet. The pacemaker stuff is really only
necessary if you want to work towards not losing any messages in the
event of a failure.

Matthew