Message loss after retry max-times

[FranklinSun]

Hi,

I use Dead Letter Exchange(DLX) and spring-retry for the bussness exceptions handling when consuming a message.

I hope, when certain exception is throwed, i will retry it. if max-retry-count is exhausted, and the message is still not be processing appropriately, i will not requeue it any more, and put it to the dead-letter-exchange.

Actual case, the message has been retry max-times, and processing still fails. But, the message has not been delivered to the dead-letter-exchange. it's gone, lost.

i don't know why? here is my xml files:

Producer:

	<rabbit:connection-factory id="connectionFactory"
		username="guest" password="guest"
		host="192.168.1.234"/>

	<rabbit:admin connection-factory="connectionFactory" />
	
	<rabbit:template id="rabbitTemplate"
		connection-factory="connectionFactory" channel-transacted="true"
		exchange="asyncservice.order.exchange" />

	<rabbit:queue name="ha.order.fujian.queue" />
	
	<rabbit:queue name="ha.order.yunan.queue" />

	<rabbit:topic-exchange name="asyncservice.order.exchange">
		<rabbit:bindings>
			<rabbit:binding queue="ha.order.fujian.queue" pattern="#.fujian" />
			<rabbit:binding queue="ha.order.yunan.queue" pattern="#.yunan" />
		</rabbit:bindings>
	</rabbit:topic-exchange>

consumer:

	<rabbit:connection-factory id="connectionFactory"
		username="guest" password="guest"
		host="192.168.1.234"/>

	<rabbit:admin connection-factory="connectionFactory" />

	<rabbit:queue name="ha.order.fujian.queue" />
	
	<rabbit:queue name="ha.order.yunan.queue" />
		
	<rabbit:queue name="ha.order.fujian.queue.dead">
		<rabbit:queue-arguments>
			<entry key="x-dead-letter-exchange" value="asyncservice.order.exchange.dead" />
		</rabbit:queue-arguments>
	</rabbit:queue>
	
	<rabbit:queue name="ha.order.yunan.queue.dead">
		<rabbit:queue-arguments>
			<entry key="x-dead-letter-exchange" value="asyncservice.order.exchange.dead" />
		</rabbit:queue-arguments>
	</rabbit:queue>

	<rabbit:topic-exchange name="asyncservice.order.exchange.dead">
		<rabbit:bindings>
			<rabbit:binding queue="ha.order.fujian.queue.dead" pattern="#.fujian" />
			<rabbit:binding queue="ha.order.yunan.queue.dead" pattern="#.yunan" />
		</rabbit:bindings>
	</rabbit:topic-exchange>
	
	<rabbit:listener-container
		connection-factory="connectionFactory" channel-transacted="true"
		transaction-manager="rabbitTransactionManager" transaction-size="1"
		prefetch="100" concurrency="10" advice-chain="retryInterceptor" >
		<rabbit:listener queues="ha.order.fujian.queue" ref="fujianMsgHandle"
			method="handleMessage" />
		<rabbit:listener queues="ha.order.yunan.queue" ref="yunanMsgHandle"
			method="handleMessage" />
	</rabbit:listener-container>
	
	<bean id="rabbitTransactionManager" class="org.springframework.amqp.rabbit.transaction.RabbitTransactionManager">
		<property name="connectionFactory" ref="connectionFactory"/>
	</bean>
	
	<bean id="retryInterceptor" class="org.springframework.amqp.rabbit.config.StatefulRetryOperationsInterceptorFactoryBean">
		<property name="messageRecoverer" ref="rejectAndDontRequeueRecoverer"/>
		<property name="retryOperations" ref="retryTemplate" />
	</bean>	
	
	<bean id="rejectAndDontRequeueRecoverer" class="org.springframework.amqp.rabbit.retry.RejectAndDontRequeueRecoverer"/>
	
	<bean id="retryTemplate" class="org.springframework.retry.support.RetryTemplate">
		<property name="backOffPolicy">
			<bean class="org.springframework.retry.backoff.ExponentialBackOffPolicy">
				<property name="initialInterval" value="5000" />
				<property name="maxInterval" value="120000" />
				<property name="multiplier" value="2" />
			</bean>		
		</property>
		<property name="retryPolicy">
			<bean class="org.springframework.retry.policy.SimpleRetryPolicy">
				<property name="maxAttempts" value="3" />
			</bean>		
		</property>
	</bean>

Am I missing a configuration setting or is it a bug?
I am using spring-amqp 1.1.3

Thanks for your help.

[Gary Russell]

You have the x-dead-lettter-exchange argument on the dead letter queue instead of the original queue.

Rabbit needs to know where to route the dead letter when the message is rejected (from the original queue). You have the DLX argument on the wrong queues, it should be on the other queues. Given that the original queues have no DLX, Rabbit silently drops the rejected message.

http://www.rabbitmq.com/dlx.html

[Wuaner]

x-dead-lettter-exchange argument should add to the original queue. And, you also need Binding a DLQ(dead-letter queue) to DLX. like this:

Both Producer & Consumer side:

<rabbit:queue name="ha.order.fujian.queue" >
  <rabbit:queue-arguments>
    <entry key="x-dead-letter-exchange" value="asyncservice.order.exchange.dead" />
    <entry key="x-dead-letter-routing-key" value="ha.order.fujian.queue.dead" />
  </rabbit:queue-arguments>
</rabbit:queue>

DLX & DLQ & Binding on Consumer side:

<rabbit:queue name="ha.order.fujian.queue.dead" />

<!-- or topic type, whatever -->
<rabbit:direct-exchange name="asyncservice.order.exchange.dead" >
  <rabbit:bindings>
    <rabbit:binding queue="ha.order.fujian.queue.dead" key="ha.order.fujian.queue.dead" />
  </rabbit:bindings>
</rabbit:direct-exchange>

[FranklinSun]

thks a lot, Gary , Wuaner

Beautiful, it's working.

Cheers.
-Franklin