[AMQ-3568] Consumer auto acking of duplicate message dispatch can lead to Unmatched acknowledge: and redelivery

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 5.5.0, 5.5.1
• Fix Version/s: 5.6.0
• Component/s: Broker, JMS client
• Labels: dispatch duplicate failover
• Assignee: Gary Tully
• Reporter: Gary Tully
• Created: 26/Oct/11 17:50
• Updated: 27/Oct/11 10:28
• Resolved: 27/Oct/11 10:28

Description

javax.jms.JMSException: Unmatched acknowledge: MessageAck {commandId = 4208, responseRequired = false, ackType = 2, consumerId = ID:gtmbp.local-35153-1319651042567-3:2:1:1, firstMessageId = ID:gtmbp.local-35153-1319651042567-3:2:1:975:2, lastMessageId = ID:gtmbp.local-35153-1319651042567-3:2:1:1050:2, destination = queue://TestQueue, transactionId = null, messageCount = 151, poisonCause = null}; Expected message count (151) differs from count in dispatched-list (152)
	at org.apache.activemq.broker.region.PrefetchSubscription.assertAckMatchesDispatched(PrefetchSubscription.java:455)
	at org.apache.activemq.broker.region.PrefetchSubscription.acknowledge(PrefetchSubscription.java:206)
	at org.apache.activemq.broker.region.AbstractRegion.acknowledge(AbstractRegion.java:427)
	at org.apache.activemq.broker.region.RegionBroker.acknowledge(RegionBroker.java:569)
	at org.apache.activemq.broker.BrokerFilter.acknowledge(BrokerFilter.java:77)
	at org.apache.activemq.broker.TransactionBroker.acknowledge(TransactionBroker.java:276)
	at org.apache.activemq.broker.MutableBrokerFilter.acknowledge(MutableBrokerFilter.java:87)
	at org.apache.activemq.broker.MutableBrokerFilter.acknowledge(MutableBrokerFilter.java:87)
	at org.apache.activemq.broker.TransportConnection.processMessageAck(TransportConnection.java:477)
	at org.apache.activemq.command.MessageAck.visit(MessageAck.java:229)
	at org.apache.activemq.broker.TransportConnection.service(TransportConnection.java:318)
	at org.apache.activemq.broker.TransportConnection$1.onCommand(TransportConnection.java:181)
	at org.apache.activemq.transport.MutexTransport.onCommand(MutexTransport.java:50)
	at org.apache.activemq.transport.WireFormatNegotiator.onCommand(WireFormatNegotiator.java:113)
	at org.apache.activemq.transport.AbstractInactivityMonitor.onCommand(AbstractInactivityMonitor.java:229)
	at org.apache.activemq.transport.TransportSupport.doConsume(TransportSupport.java:83)
	at org.apache.activemq.transport.tcp.TcpTransport.doRun(TcpTransport.java:222)

Problem occurs when a duplicate dispatch is one of many inflight messages to a destination. The duplicate detection auto acks with a standard ack in place of an individual ack. The standard ack results in an exception in this case because it does not match the dispatch list of the broker. optimizeAcknowledge on the connection factory seems to make this more probable. The duplicate originates from a failover: recovery/reconnect resend.

The end result is pending messages on the queue and redelivery after a restart.

In some cases, the need for duplicate detection can be circumvented at source via the kahaDB store producer audit, the default LRU cache size is 64. Increasing this can help.

<kahaDB ... maxFailoverProducersToTrack="2048" />