I am using JMS connection pool and concurrent consumers (with the randomize=true setting in failover uri). I am not sure if it will cause the "consumer bounces between brokers" scenario you described. If so, does that means my other messages may also be stuck on a broker?
I tried to use the following option in the activemq.xml setting as the activemq site suggested ( http://activemq.apache.org/networks-of-brokers.html#NetworksofBrokers-StuckMessages%28version5.6%29 <http://activemq.apache.org/networks-of-brokers.html#NetworksofBrokers-StuckMessages%28version5.6%29> ) but activemq had problem even load it (I am using activemq 5.7): Caused by: org.xml.sax.SAXParseException: cvc-complex-type.2.4.a: Invalid content was found starting with element 'conditionalNetworkBridgeFilterFactory'. One of '{"http://activemq.apache.org/schema/core":deadLetterStrategy, "http://activemq.apache.org/schema/core":destination, "http://activemq.apache.org/schema/core":dispatchPolicy, "http://activemq.apache.org/schema/core":messageEvictionStrategy, "http://activemq.apache.org/schema/core":messageGroupMapFactory, "http://activemq.apache.org/schema/core":networkBridgeFilterFactory, "http://activemq.apache.org/schema/core":pendingDurableSubscriberPolicy, "http://activemq.apache.org/schema/core":pendingMessageLimitStrategy, "http://activemq.apache.org/schema/core":pendingQueuePolicy, "http://activemq.apache.org/schema/core":pendingSubscriberPolicy, "http://activemq.apache.org/schema/core":slowConsumerStrategy, "http://activemq.apache.org/schema/core":subscriptionRecoveryPolicy, WC[##other:"http://activemq.apache.org/schema/core"]}' is expected. gtully wrote > the dlq should behave like any other queue w.r.t a network. > One thing to note however, it that a message will only be bridged > once, so if a message is forwarded from A->B it won't by default get > forwarded back from B->A. > > I wonder if that is is occurring in your scenario. It could be if your > consumer bounces between brokers, because it will create demand and > suck messages across the bridge even if it does not consume them all. > > some more detail at: > http://activemq.apache.org/networks-of-brokers.html#NetworksofBrokers-StuckMessages%28version5.6%29 > > > On 12 November 2012 18:21, jiunjiunma < > jiunjiunma@ > > wrote: >> I had a small test case that cleans up the ActiveMQ DLQ. It works fine >> when I >> am running a single broker. When I ran a network of brokers (2 static >> brokers using per queue based DLQ setting), I noticed the client code >> (with >> uri=failover://(tcp://localhost:61616,tcp://localhost:61618)?randomize=true) >> only remove the messages from the instance DLQ it connected to. It seems >> the >> client consumer was only able to access the DQL the activemq instance is >> current connected to and if the message happened to be on the other >> broker >> then it was not able to be deleted. >> >> It was a bit surprise to me because I would expect the DLQ should be >> behaved >> the same as the normal queues and messages should be processed by the >> consumer regardless of the broker it connected to. Is this the expected >> behavior? Or my understanding was wrong that the DLQ messages will not be >> store-and-forwarded to the other brokers? >> >> I'd appreciate it if anybody can shed some light on it. >> >> Thanks, >> --Jiunjiun >> >> >> >> -- >> View this message in context: >> http://activemq.2283324.n4.nabble.com/Will-messages-in-DLQ-be-store-forwarded-in-network-of-broker-setting-tp4659196.html >> Sent from the ActiveMQ - User mailing list archive at Nabble.com. > > > > -- > http://redhat.com > http://blog.garytully.com -- View this message in context: http://activemq.2283324.n4.nabble.com/Will-messages-in-DLQ-be-store-forwarded-in-network-of-broker-setting-tp4659196p4659212.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.