[rabbitmq-discuss] rabbitmq restarted in bad state

[Simon MacMullen]

On 15/02/12 01:37, Santosh Thammana wrote:
> Hello,
>
> I had some issues with RabbitMQ, and had to restart it.  The issue was
> that the queue sizes were remaining constant, despite there being
> producers and consumers.  But after restarting, I am unable to create a
> few of my queues.
>
> I am using Pika, and I get the error:
> ChannelClosed: Channel.Close(class_id = 50, method_id = 10, reply_code =
> 404, reply_text = "NOT_FOUND - no queue 'collector.be2' in vhost '/'")

The only way to get a NOT_FOUND when declaring a queue is if the queue 
has previously been declared on a cluster node which is currently down. 
Does that describe your situation?

> I get the same error if I then try to delete it.
> We noticed that the queue is still bound to the default exchange.  So,
> it seems like that's causing the strange behavior.  And I can't delete a
> default binding.

The default binding should not prevent you from declaring the queue.

>  Is there any way to solve this problem without
> restarting RabbitMQ?  If I do have to restart RabbitMQ, can I expect it
> to be in a proper state?  I noticed that at least one of the queues I
> had issues with was declared as durable.  I'm not sure if that impacted
> this negatively.

I think they would all have to be durable.

Cheers, Simon

-- 
Simon MacMullen
RabbitMQ, VMware