Further investigation reveals, that the message id's that it is complaining
about not being able to recover, are actually waiting now successfully in
another queue....

That is, my application submits 12 messages to queue1, then a consumer
consumes those messages and processes them, and sends them to queueB.  After
this, I look at the webConsole, and see that queueA has 0 Pending messages,
and queueB has 12.

Then I restart the broker, and on restart, the broker log shows 8 messages
that it complains about not being able to recover.  The webConsole now shows
8 pending messages for queueA, and still 12 for queueB.  The 8 messageIds
that it says it can't recover, match id's that are now sitting in queueB. 
If I browse the queues, it shows no messages in queueA, and the 12 messages
in queueB look good, including the 8 of those that match id's that it
complained about.

Finally, if I restart the broker subsequently, it again complains about not
being able to recover 8 messages....('already dispatched')....

Jason



Jason Rosenberg wrote:
> 
> Ok,
> 
> I enabled DEBUG logging on the broker, and now I see this on restart:
> 
> 2008-02-10 03:52:43,645 [127.0.0.1:64511] DEBUG RecoveryListenerAdapter       
> - Message id ID:szjrosenberg-mbp17.local-64382-1202633445267-0:22:1:1:1
> could not be recovered from the data store - already dispatched
> 
> In fact, I see 8 of these, and the webconsole in this case now shows '8'
> pending messages for the queue, but never delivers them.  The webconsole
> showed 0 pending messages before shutting down, etc...
> 
> I guess there are 2 issues.  First, if it can't recover the message, it
> probably shouldn't show it as pending.   Second, need to find out why it's
> even trying to recover the message in the first place, after showing no
> Pending message before shutdown.
> 
> I'm using a 5.1-SNAPSHOT broker that I downloaded around Jan 24, 2008....
> 
> Jason
> 
> 
> 
> Jason Rosenberg wrote:
>> 
>> Now, after closing down all connected clients, and restarting the broker,
>> the webconsole shows the same queue has 2 pending messages....Clicking on
>> browse messages shows no messages.....
>> 
>> I'm wondering if the jmx QueueViewMBean needs to have more refinement to
>> display more detailed status of messages that are in various states (e.g.
>> cached, redelivered, recovered, rolled-back, etc.)...
>> 
>> Or is it likely a bug altogether...
>> 
>> Fun stuff
>> 
>> Jason
>> 
>> 
>> 
>> Jason Rosenberg wrote:
>>> 
>>> I just tried clicking on "purge" in the webconsole, on a queue that
>>> seemed to have messages stuck in the pending state.  After the purge, it
>>> now shows that it has -4 messges pending.
>>> 
>>> Should it ever be a negative number of messages pending?
>>> 
>>> Jason
>>> 
>>> 
>>> 
>>> 
>>> Jason Rosenberg wrote:
>>>> 
>>>> I have a scenario, with amq 5.1....
>>>> 
>>>> Where more messages are being received than are sent.  Should that be
>>>> possible?  It looks like indeed some of the messages are getting
>>>> delivered twice, and tallied as such in the webconsole view of the
>>>> queue....Is there some setting I need to place on my queues to make
>>>> sure this doesn't happen?  Or is it a bug...
>>>> 
>>>> I'm using the amqPersistence for persistent messaging.
>>>> 
>>>> Also, see my previous post, regarding the broker showing new messages
>>>> Pending after restarting the broker....
>>>> 
>>>> Weirdness, it would seem....
>>>> 
>>>> The config is very much the same as what I had previously with 4.1.1,
>>>> using the default persistence there (journaled file-based jdbc)....
>>>> 
>>>> Jason
>>>> 
>>> 
>>> 
>> 
>> 
> 
> 

-- 
View this message in context: 
http://www.nabble.com/Sent-%2B-Pending-should-equal-Received%2C-no--tp15367886s2354p15394870.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Reply via email to