We have an application which has an EJB which runs Asyn and is annotated as @TransactionAttribute(TransactionAttributeType.REQUIRES_NEW) (I believe this is default behavior but for clarity we have added). When running the application we keep getting the following:

 

12:54:53,696 WARN  [com.arjuna.ats.arjuna] (Transaction Reaper) ARJUNA012117: TransactionReaper::check timeout for TX 0:ffffac124a3d:598c1ee6:502bc969:51 in state  RUN

 

 

This is normal behaviour when the timeout period of your transaction has expired. Try increasing the timeout or decrease the amount of work you perform in your transaction.

Yes that should have worked (but you mean one of the standalone.xml config files - did you change the right one though).

Could you see if setting the timeout via the cli makes any difference:

bin/jboss-cli.sh --connect

[standalone@localhost:9999 /] /subsystem=transactions/:write-attribute(name=default-timeout,value=10000)

and then read it back to check that it worked:

[standalone@localhost:9999 /] /subsystem=transactions/:read-attribute(name=default-timeout)

and then reboot the server

Yeah sure but is your transaction still timing out prematurely.

Could you produce a simple test case. For example, I had a go at reproducing your issue using the http://www.jboss.org/jdf/quickstarts/jboss-as-quickstart/cmt/ AS7 quickstart. I modified src/main/java/org/jboss/as/quickstarts/cmt/ejb/CustomerManagerEJB.java to match what I think you have as follows:

@javax.ejb.Asynchronous

@TransactionAttribute(TransactionAttributeType.REQUIRES_NEW)

void createCustomer(String name) throws RemoteException, JMSException {

    try {


             Thread.sleep(2000);

         } catch (

    etc

}


The 1 sec sleep is to give time for the transaction to timeout. Then deploy your changes, change the default timeout via the CLI (or standalone-full.xml)

   /subsystem=transactions:write-attribute(name=default-timeout,value=1)

and then reboot the server.

Now when you invoke this ejb method you should see the transaction timing out and aborting. Look for something similar to the following lines on the console:

14:37:02,172 ERROR [org.jboss.ejb3.invocation] (EJB default - 2) JBAS014134: EJB Invocation failed on component CustomerManagerEJB for method public void org.jboss.as.quickstarts.cmt.ejb.CustomerManagerEJB.createCustomer(java.lang.String) throws java.rmi.RemoteException,javax.jms.JMSException: javax.ejb.EJBTransactionRolledbackException: Transaction rolled back

 

....

 

14:37:02,182 ERROR [org.jboss.as.ejb3] (EJB default - 2) JBAS014102: Asynchronous invocation failed: javax.ejb.EJBTransactionRolledbackException: Transaction rolled back

If you set the timeout back to the default you will see the transaction commit successfully, thus proving that the default-timeout property works as it should.

I made the changes to the quickstart for real and the default-timeout property works as I outlined in my previous post.

So my test case shows that the transaction manager respects the default-timeout property. If you want to see the proof simply repeat my instructions by reading my previous post.

But as I said in the post "I modified CustomerManagerEJB.java to match what I think you have" so it is for you to decide if my test case matches what you are doing in your own application.

I had a similar problem and increasing the timeout as suggested above

bin/jboss-cli.sh --connect

[standalone@localhost:9999 /] /subsystem=transactions/:write-attribute(name=default-timeout,value=10000)

1.         <subsystem xmlns="urn:jboss:domain:transactions:1.1"> 
2.             <core-environment> 
3.                 <process-id> 
4.                     <uuid/> 
5.                 </process-id> 
6.             </core-environment> 
7.             <recovery-environment socket-binding="txn-recovery-environment" status-socket-binding="txn-status-manager"/> 
8.             <coordinator-environment default-timeout="10000"/> 
9.         </subsystem>   

Worked fine