Hello Friends, i am trying to develop a program where i put a message in a request queue, from the request queue mdb picks up the message, does some processing and puts the response back in response queue. The calling program tries to read the message from the response queue using QueueReciever and JMSCorrelation Id. But some how things are not working right. My QueueReceiver is not able to pick the message from the response queue.
Following is the code written in the SessionBean which is sending and recieving message.
Following is the code written in MDB
Plz help i am unable to understand where i am wrong. I am using WSAD 5.1.2 Developer Edition on Windows. [ July 02, 2005: Message edited by: Prakash Dwivedi ]
Prakash Dwivedi (SCJP2, SCWCD, SCBCD)
"Failure is not when you fall down, Its only when you don't get up again"
The code seems ok. I'm guessing the problem is the same one I used to have way back. Have you tried adding some debugging statements to your code? If you put them in the correct places I think you might see something not quite unlike the following scenario: - Session bean: Sending message - Session bean: Waiting for reply - Session bean: Timed out waiting for reply - MDB: Received message - MDB: Sending reply - MDB: Reply sent
What happens here is that both the send and the receive in your test code are done in the same transaction. If the send method does not throw an exception, that only means that the transport layer has accepted your message, not that it has actually been sent. (The mail is in the mailbox). For the message to be sent, the current transaction has to complete succesfully. Which is done (at the earliest) after sendRecvMessage is finished.
The easiest way to solve this problem and make your test case work, is to give the receive (getMessage) method a transaction attribute of RequiresNew. This will start a new transaction, closing the previous one, allowing the message to be send.
Some constructive criticism: (or: things you didn't ask for ) In stead of looping like: for(int i=0;i<100000;i++); to wait for a small period, use: Thread.sleep( 100 ); // in milliseconds
This will not cost cpu time. The looping for time is really ASM/Pascal age
Try closing things in the opposite order of opening: Open connection Open session Close session Close connection
This may avoid unexpected exceptions and generally is considered good coding practice.
currently i am doing websphere migraiton from 4.0 to 5.1.0
i am also facing the same problem what you have
A stateless session bean puts a message in a request queue, From the request queue a C++ Program (independent component) picks up the message, does some processing and Puts the response back in reply queue. The session bean then tries to read the message from the reply queue using QueueReciever and JMSCorrelationID in a separate JMS Session. But QueueReceiver is not able to pick the message from the reply queue. Using createBrowse I am able to read the messages in the reply queue.
Observations: 1) Same code is working in Websphere 4.0, not working in websphere 5.1.0 3) I created one jsp which have JMS code to get the message using JMSCorrelationID. It is working fine. Note: it is reading the message from web container not from EJB Container
Q. Why can't I receive a message that I send within a container-managed transaction?
A. If you are using container-managed transactions, the original message sent from the EJB will never be sent. Here is what is happening.
Container starts transaction.
Generate new message.
Send message (message isn't sent - it's buffered until transaction commit).
Do a blocking receive on a queue.
Transaction Commit never Reached because original message was never sent because you can't get past blocking receive. The solution is to either use bean-managed transactions, or to break the send and receive into two separate methods.
The solution is to either use bean-managed transactions, or to break the send and receive into two separate methods.
And then set a REQUIRESNEW transaction attribute on the receive method!
Oh and one other fun detail I found out after we deployed the application on WebSphere on AIX. Specifying a negative timeout (bug in my own round robin receive algoritm) on Windows would result in an IllegalArgumentException, whilst on AIX it would be interpreted as 0 (block indefinitely). A minor but _very_ annoying detail.
Quis custodiet ipsos custodes?
Every snowflake is perfect and unique. And every snowflake contains a very tiny ad.