[rabbitmq-discuss] Resource alarms in 3.1 seems to block operations other than basic.publish

Mon May 13 18:00:58 BST 2013

On 13 May 2013, at 17:02, Michael Klishin wrote:

> 2013/5/13 Tim Watson <tim at rabbitmq.com>
> 1. are you sure the queue is not empty by the time the consumer is restarted?
> 
> I typically restart the publisher first, so yes, there are certainly messages in the queue

I didn't try that, but it wouldn't make any difference if that alarm was still set, since the producer would get blocked immediately would it not?

>  
> 2. do you receive no methods at all in the consumer, not even connection negotiation?
> 
> Correct. But it appears to be that any network activity almost locks up, I blame the OS X kernel that's very easy to
> overwhelm with flood publishing on localhost.

I've certainly been unable to replicate this across two machines, and with just localhost on Snow Leopard, I never get the queue to be anything but empty, that is once the producer is blocked and the consumer is receiving no more messages, running `rabbitmqctl list_queues' shows a queue length of 0. How do you get more messages into the queue then, if a memory alarm is still set? I didn't think that was possible?

Cheers,
Tim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130513/5f6910c1/attachment.htm>