WebLogic JMS Production Timeout

[Dave Barton]

Hi,

We are using WLS 7.0 SP1 and SP4 and have experienced this problem on
both.

A JMS client periodically gets this exception:

1: weblogic.jms.common.JMSException: ; nested exception is:
java.io.EOFException
weblogic.jms.common.JMSException: ; nested exception is:
java.io.EOFException
at weblogic.jms.dispatcher.DispatcherWrapperState.dispatchSyncTran(DispatcherWrapperState.java:333)
at weblogic.jms.client.JMSProducer._send(JMSProducer.java(Compiled
Code))
at com.ft.info.util.messaging.gp3.PTPMessaging.produceMessage(PTPMessaging.java(Compiled
Code))
at com.ft.info.util.messaging.gp3.PTPMessaging.produceMessage(PTPMessaging.java(Compiled
Code))
at com.ft.info.util.messaging.gp3.MessagingProducerClient.produceMessage(MessagingProducerClient.java(Compiled
Code))
at com.ft.info.gp3.addb.workflow.ADDBDeliveryRequestor.sendRequest(ADDBDeliveryRequestor.java(Compiled
Code))
at com.ft.info.gp3.addb.daemon.DeliveryRequest.produce(DeliveryRequest.java(Compiled
Code))
at com.ft.info.gp3.addb.daemon.DeliveryRequest.run(DeliveryRequest.java:182)
at java.lang.Thread.run(Thread.java:512)
----------- Linked Exception -----------
weblogic.rjvm.PeerGoneException: ; nested exception is:
java.io.EOFException
java.io.EOFException
at weblogic.rjvm.t3.T3JVMConnection.endOfStream(T3JVMConnection.java:650)
at weblogic.socket.JavaSocketMuxer.processSockets(JavaSocketMuxer.java:331)
at weblogic.socket.SocketReaderRequest.execute(SocketReaderRequest.java:23)
at weblogic.kernel.ExecuteThread.execute(ExecuteThread.java:153)
at weblogic.kernel.ExecuteThread.run(ExecuteThread.java:134)

We have eventually narrowed this down to Text messages of 200K and
over. The failure happened about 1 minute after the production of the
message started which led us to the CompleteT3MessageTimeout WLS
config parameter. Increasing this to 2 minutes and then 3 minutes led
to corresponding increases in the time taken for the JMS connection to
fail. Eventually a value of 4 minutes got rid of the failure.

My question is what would cause the production of a 200K Text message
to take more than 3 minutes? The client and WL server are on different
nodes but talk across a fast internal network. The JMS queue is
persisted to an Oracle database, could this is taking up the time?

Thanks,
Dave Barton
FT Interactive Data
London