[rabbitmq-discuss] Possible disconnected queue?

[Austin Buckner]

Thanks! Deleting that directory cleared up our issue of having to recreate
a queue with the same name.

But I agree this has happened once so it will probably happen again.

The dirs are nfs mounted.

-Quinn

On 8/24/12 11:55 AM, "Matthias Radestock" <matthias at rabbitmq.com> wrote:

>Austin,
>
>On 24/08/12 16:48, Austin Buckner wrote:
>> # ls -ltaR
>>
>>/mnt/netapp12/rabbitmq_test/mnesia/rabbit\@diskrabbit01-test/queues/2PQ75
>>8A
>> TP80LLMCCE806NG6JK/
>>
>>/mnt/netapp12/rabbitmq_test/mnesia/rabbit at diskrabbit01-test/queues/2PQ758
>>AT
>> P80LLMCCE806NG6JK/:
>> total 20
>> drwxr-xr-x 136 rabbitmq rabbitmq 16384 Aug 24 10:43 ..
>> drwxr-xr-x   2 rabbitmq rabbitmq  4096 Aug 23 13:14 .
>>
>>
>> Looks like the only difference between this dir and the rest is that its
>> missing the journal.jif file
>
>Curious. So one plausible sequence of events is this:
>
>1. queue gets deleted
>2. rabbit attempts a recursive queue deletion by...
>2.1 getting the dir listing
>2.2 deleting all the files
>2.3 deleting the dir - bang! the dir still has files in it
>3. some time later the remaining file(s) in the dir vanish
>
>I wonder whether there is some peculiar filesystem issue here. Noting
>the /mnt prefix, are these dirs nfs mounted or on some other sort of
>network storage?
>
>
>To get you going again, it may be sufficient to delete the dir manually;
>you might then be able to re-declare a queue with the same name.
>
>But given that the fault occurred before, it's likely going to occur
>again.
>
>
>Regards,
>
>Matthias.
>