[rabbitmq-discuss] queue_declare returning a (50, 10) NOT_FOUND....

[Matthias Radestock]

Nicolas,

Nicolás César wrote:
> No. this is my declaration
> 
>         self.chan_in.queue_declare(queue=self.config_in['queue'],
>                                    durable=True,
>                                    exclusive=True,
>                                    auto_delete=False)
> 
> 
> but there is a strange issue, if I change the name... to something I 
> havent used, there is no problem; just like I did in
> http://groups.google.com/group/rabbitmq-discuss/browse_thread/thread/2d925b0bbaed7f86?pli=1
> 
> is there a presistent queue "namespace" table that maybe got corrupted? 
> BTW is one  of the few "exclusive" queues that I have. maybe a wrong 
> consumer connected stored somewhere?

Exclusive queues only survive for as long as the connection that created 
them. (As an aside, this also means declaring an exclusive queue as 
durable is somewhat pointless.) And an exclusivity violation gets 
reported as different error than NOT_FOUND. So no, that's not it.

As Simon mentioned, the only case in which an active queue.declare 
should report a NOT_FOUND error is when it encounters a durable queue 
residing on a cluster node that is currently down. Now, your 
'rabbitmqctl status' output indicates that the system isn't clustered. 
But from your previous posts I gather that perhaps it was clustered 
earlier. It is possible that the unclustering left something behind. If 
the above error is reproducible, please let us know.

> I've solved this changing the name (just like before), so it's over for me.,

Next time this happens, list the existing queues with 'rabbitmqctl 
list_queues name durable auto_delete pid owner_pid' and post the result 
here.

Regards,

Matthias.