[rabbitmq-discuss] General questions about HA, Stability/Reliability and Broker Administration

Mon Jul 26 11:32:11 BST 2010

Hi Dave. I'll just reply to a couple of your points as well; hopefully 
eventually you'll get a full set of answers.

On 24/07/10 16:33, Dave Greggory wrote:
> 4. What tools do you use for broker administration in production environments?
> Does anyone use BQL and have a good experience running it without any issues in
> a high volume/guaranteed delivery environment? It's not yet officially released
> so I don't want to run it in a production environment. If not, how do you
> manage? Tools built in-house? Is it even possible to do that without resorting
> to programming Erlang?

Depends what you mean by broker administration. RabbitMQ provides 
everything you need in theory out of the box: use rabbitmqctl for user / 
vhost / permission administration and get your applications to declare 
all the AMQPish resources they need.

Some people prefer a unified view and a central point of administration 
for these things; at the moment BQL and Alice are the best options 
available.

> 5. How do you monitor RabbitMQ nodes and what do you monitor for (I'm thinking
> of monitoring various queue sizes)? The Zenoss AMQP Plugin looks interesting to
> us, but I have been informed that it's not playing nice with latest Zenoss.

There's a variety of monitoring tools people have put together, but as 
far as I know they're all really doing the equivalent of "rabbitmqctl 
list_queues" and exporting that somewhere. So all you get is a way to 
check the number of messages in queues (and polling that goes and wakes 
all your queues up).

We're aware that our monitoring situation is quite poor (and our 
management situation could use some polish) and we (well, I) am starting 
to take steps to improve the situation. Watch this space.

Cheers, Simon
-- 
Simon MacMullen
Staff Engineer, RabbitMQ
SpringSource, a division of VMware