I would like to ask experienced Spring Integration developers or even code authors regarding weird behavior Im experiencing with my code. The issue is that Spring Code executes COMMIT on my RuntimeException. So here is my setup:
1.
I have standard JMS inbound channel adapter configured using
<jms:message-driven-channel-adapter />
I have transactionManager which implements the Best effort 1PC pattern - that one available in Spring, ChainedTransactionManager.
Im consuming messages in simple testing flow where the JMS inbound adapter only publishes to direct channel and via one transformer endpoint the converted message is published to another direct endpoint where the handler which is ServiceActivator intentionally throws my application exception which extends RuntimeException.
Now when I debug the whole flow and when Im stepping through the Spring internal code I always end in place where the configured transaction manager calls COMMIT and thus my ChainedTransactionManager executes COMMIT as well despite Im simulating error and thus clear rollback scenario. The breakpoint in the rollback method is not hit at all.
What can I have wrong?
Now I have gone through the code again and even the Spring code sets the status to RollbackOnly and executes some exception handlers but at the end it still goes on successfully and ultimately it calls the commit line as shown on the screenshot and chainedTransactionManager instructs its local TMs to commit as well.
Im hoping there is some detail/mistake Im not able to spot :) Thanks!
Screenshot from class .m2\repository\org\springframework\spring-jms\5.2.3.RELEASE\spring-jms-5.2.3.RELEASE-sources.jar!\org\springframework\jms\listener\AbstractPollingMessageListenerContainer.java
Related
I have recently had to support a colleague in verifying why some system tests are not passing in wildfly, system tests that pass consistently on weblogic and glass fish.
After analysing the log, it became clear the reason is related to a JMS message sent by a backed thread getting committed to a queue too soon, when the expectation was the message would be committed when the entry point Container Managed Transaction of an MDB commits. So the message is going out before the MDB that sends it is done running.
In weblogic, to achieve the expected behaviour, you need to make sure that when you take the connection factory given by the container , which is XA configured, you set the connection.createseesion with
transacted = true and
acknowledgement = session transacted.
In a process similar to the one depicted in this URL
http://www.mastertheboss.com/jboss-server/jboss-jms/sending-jms-messages-over-xa-with-wildfly-jboss-as
Except in the snippet above auto acknowledge is set and the first parameter is set to false.
In wildly when our weblogic and glass fish configuration is used, nothing is committed and the system behaves as if the JMS message sent were to be rolled back.
If configuration as in the example above were to be used, instead what would happen is that the JMS message is immediately and the consumer MDB immediately launches being trigerred before the producer transaction actually ends, causing the system test to fail.
According to the official JMS configuration, by using a connection-pooled factory with the transaction=XA attribute, the container should immediately bind the commit of the transaction to the lifecycle of the parent transaction.
See official documentation bellow in particular in respect to the Java:/JmsXa connection factory.
https://docs.jboss.org/author/display/WFLY10/Messaging+configuration
My colleague was initially using a non pooled connection factory, but the injection info reference has since then been fixed. I have tried all possible combinations of parameters in the shed message, but my outcome is sitll:
Either sent too soon or never sent.
To conclude all the other resources are XA. Namely the oracle db is using the XA driver.
Can anyone confirm if in wildly the send JMS message only when parent transaction commits is working and if so how the session is being configured?
I will check if perhaps my colleague has not made a mistake in terms of the configuration of the connection factory used by the Men's themselves to consume messages out of the queue.but if that one is also XA... Then it is a big problem.
So the issues is fixed.
The commit of the JMS message to the queue at the end of the transaction works perfectly.
The issue was two fold:
(a) First spot of code I was looking at address the issue was not correct. Someone had decided to write his own send telegram to queue API elsewhere, and was not using the central API for sneding telegrams, so any modification I to the injection connection factory was actually not taking effect. The stale connection factories were still being used.
(b) Once the correct API was spotted it was easy to make the mechanism work by using the widlfy XA pooled connection factory mentioned in the post above.
The one thing that was tweaked was the connection.CreationSession api.
The API in JEE 7 has been enlarged and it is now better documented than in jEE 6.
To send a JMS message in a container as part of an XA transaction one should do:
connection.createSession() without any parameters.
This can easily be seen in the connection javadoc:
https://docs.oracle.com/javaee/7/api/javax/jms/Connection.html
QUOTE 1:
This method has been superseded by the method createSession(int
sessionMode) which specifies the same information using a single
argument, and by the method createSession() which is for use in a Java
EE JTA transaction. Applications should consider using those methods
instead of this one.
QUOTE 2:
In a Java EE web or EJB container, when there is an active JTA
transaction in progress:
Both arguments transacted and acknowledgeMode are ignored. The session will participate in the JTA transaction and will be committed
or rolled back when that transaction is committed or rolled back, not
by calling the session's commit or rollback methods. Since both
arguments are ignored, developers are recommended to use
createSession(), which has no arguments, instead of this method.
Which means, the code snippet in:
http://www.mastertheboss.com/jboss-server/jboss-jms/sending-jms-messages-over-xa-with-wildfly-jboss-as
Is not appropriate. What one should be doing is creating the session without any parameter and leting the container handle the rest.
Which it does just fine.
In Spring based application, Transaction Manager is responsible for committing or rolling back SQL transactions.
My application uses a custom cache for some part of persistent data.
This cache is not managed by Spring nor Hibernate.
If a SQL transaction encounters errors and must be rolled back, then cache modifications should be rolled back as well.
My question is, how to register an event listener or callback which will call my cache.evict() method when Spring Transaction Manager rolls back a transaction?
You can use TransactionSynchronizationAdapter for this. For reference, you can look at this: Spring - commit JMS Transaction after JPA transaction
In the answer given in the link, the synchronization is registered for afterCommit. In your case, it would be afterCompletion. In that implementation, you will need to check if the transaction is in a rolled back state and do the cache eviction.
There are lots of ways of doing this...
You can use application events if you want... Application Events in Spring Framework 4.2
You can throw a custom runtime exception and you can handle it in your exception handler (if you're using spring MVC). but here I don't recommend to do any important operations like clearing of cache here... Exception handling in Spring MVC
You can use a combination of #1 and #2. You can send an event that will eventually throw a runtime exception that you handle it in UI (spring mvc or whatever you use). This is how I would do it
You can throw an exception and anyone calling your bean will get the custom transaction exception you want and will have to deal with notification... I don't recommend this
I am working on a project which makes extensive use of spring transactions. I t so happened that i was throwing an exception AND NOT HANDLING IT PROPERLY which does not commit or rollback the transact Ion. So the connection remains active even when the thread is stopped. When new request to the web server ( Apache tomcat 7.0) comes the spring provides the earlier connection to the new thread. Since the thread didn't started the connection so the thread couldn't close it either ie .commit doesn't work. because of which the objects are not getting persisted in database in the consequent transactions even when there is no exception. How can i work around the problem so that i can detect where is the actual problem happening or can a design an exit point where i can explicitly close the transaction before response is sent to browser.
A runtime unchecked exception or error that is thrown from a #Transactional method will rollback that transaction by default; a return or a checked exception will commit the transaction by default. You can tell Spring to override that behaviour using the annotation type elements of the #Transactional annotation.
The Spring Framework reference manual has an entire chapter devoted to transaction management. You need to read and understand that as your starting point.
I know that Camel's JMS component, for receiving messages, uses Springs DefaultMessageListenerContainer. It can be configured to use CLIENT_ACKNOWLEDGE mode for acknowledging messages. My question is, when exactly the message.acknowledge() method gets called? Is it called internally by the spring's listener container?
Or can I somehow acknowledge message at my will?
I'd like to avoid scenario that messages get lost because my app crashed during processing of these messages and making it transactional seems a bit too heavy for me
OK. After some some debugging and scanning throug source code I've found out that Camel uses spring MessageListenerContainers. The AbstractMessageListenerContainer, in case of CLIENT_AKNOWLEDGE mode, invokes comitIfNecessary method acknowledging message. This happens only AFTER registered MessageListener processes message succesfully (no exceptions)
Camel uses EndpointMessageListener which, eventually, invokes process method of next processor (or Producer) down the route. As this is classic chain of responsibilities, if any processor down the route throws exception or sets exception on the Exchange, it will be re-thrown by EndpointMessageListener preventing the AbstractMessageListener to acknowledge message.
Well, actually JBoss does send the message, but only when the current transaction started by the EJB is finished.
We have this problem with JBoss 4.0.3 and Spring's JmsTemplate. An EJB sends a message to a queue with a temporary queue as the reply_to field. Then, inside the same transaction, we listen for the response given by the first MDB. The problem is that the JmsTemplate's method "send" isn't executed after the transaction have finished. So, by the time the message is sent to the queue, and processed by the MDB, the listener of the temporary queue is gone.
This is called "Synchronous Reception"
Two things change this behavior but does raise some concerns:
Change the EJB's transaction type to BMT. (Concern: BMT sucks)
Create a thread that all it does is to call the JmsTemplate.send() method.
As a side note, this is an EJB that is working correctly on a weblogic environment, and the message does get sent when it should, in the middle of the transaction not when it's over.
Thanks for any help.
JBoss's behaviour is correct. JMS is a transactional API, and sends should only be executed when the tx commits.
It may be possible to convince JmsTemplate not use the current transactional context, although it makes a point of trying to hide the unpleasantness of the JMS API from you.
You could wrap the JMS template in code, either a Stateless session bean or a service method using Spring's transaction management, that uses a transaction propagation of REQUIRES_NEW. That way the sending of the message is in its own transaction that will commit the sending of the message outside the scope of the wrapping transaction.
I'm not sure why this would work on Weblogic though. My guess would be that on Weblogic it's not recognizing the queue as an XA Queue.