Redelivery does intentionally block the queue to maintain order and when
messages are sent to the dlq, they do indeed stay there.

You can check out a camel based solution to redelivery processing from
https://issues.apache.org/activemq/browse/AMQ-2710 (see the subversion
commit tab to see the source). You can build the delay into camel pending
the scheduled delivery feature in 5.4.

The use of a composite queue is an interesting alternative approach but you
may be running into duplicate suppression, not sure, would need a little
test case to investigate further.

On 25 May 2010 02:35, sic <sic_1...@naver.com> wrote:

>
> I'm using 5.2.0 and transfer data with queue.
>
> In this state, when consumer failed to get messages from the broker, the
> messages are moved by individualDeadLetterStrategy.
> here is my broker configuration:
>
>    <amq:policyEntries>
>      <amq:policyEntry queue=">" memoryLimit="50mb">
>        <amq:deadLetterStrategy>
>            <amq:individualDeadLetterStrategy
>            queuePrefix="DLQ." useQueueForQueueMessages="true"/>
>        </amq:deadLetterStrategy>
>     ...
>
> But I hope that failed messages aren't just staying DLQ, they are moved to
> the end of the original queue and redelivered to the consumer. Having said
> that, failed messages cannot be continuously stuck to the first of the
> original queue in order to process other messages normally(i.e. I want to
> send failed messages to the end of the original queue).
> fail to consume -> DLQ -> move to the end of original queue -> <reprocess>
> -> <success>
> so I try using a composite queue.
> <amq:virtualDestinations>
>  <amq:compositeQueue name="DLQ.FOO">
>        <amq:forwardTo>
>          <amq:queue physicalName="FOO" />
>        </amq:forwardTo>
>  </amq:compositeQueue>
> </amq:virtualDestinations>
>
> Despite my expectation, messages moved to the original queue can't be
> reprocessed by any consumers no longer, just seem to increase inflight
> count.
>
> Any misconception about the DLQ handling for me or alternative ways?
>
> appreciate any comments about this subject.
>
> regards,
>
> sic
> --
> View this message in context:
> http://old.nabble.com/DLQ-Handling-tp28663453p28663453.html
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>
>


-- 
http://blog.garytully.com

Open Source Integration
http://fusesource.com

Reply via email to