Open Stack

于 2014-5-6 23:11, Craig Jellick 写道:
> This happens every time we hit the configured HAProxy TCP timeout. It
> can also happen if a Rabbit node that has a connection is restarted, if
> HAProxy is restarted, or if there is basically any interruption in the
> connection between the clients and HAProxy.

Hi,

I've read  the book Rabbitmq in Action, have also setup the HAproxy for 
HA in the test environment.(wrote a blog item: 
http://www.nsbeta.info/archives/555 though it's in Chinese) For your 
case, I was thinking since the MQ consumer (the neutron agent) is 
keeping a long connection to the broker (there is also pull mode, but 
for consume mode it works that way), when the connection gets broken, 
the consumer may or may not know it. If the consumer knows, it will 
retry. otherwise it won't retry. For both cases the results have no 
guarantee. When you add a proxy between the consumer and the broker, the 
broken risk also increases. From my viewpoints (just mime), I won't 
suggest a proxy setup for RabbitMQ.

Regards.