Iron JAC Amar is being told to create more KeyConnectionAssociation under CachedConnectionManagerImpl then actual JTA transactions open (Reproducible with NESTED EJB calls) ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Key: WFLY-9542 URL: https://issues.redhat.com/browse/WFLY-9542 Project: WildFly Issue Type: Bug Components: EJB Affects Versions: 10.1.0.Final Reporter: NUNO GODINHO DE MATOS Assignee: Tomasz Adamski Priority: Major Labels: Arjuna, IronJacamar, JTA Attachments: 2020.09.03_serverLog_use_cm_set_to_false_solves_issue.7z, 2020_09_03_15_22_40_wildlfly_ironjacamar_unregisterconnection_wildfly_jta_commit_reproducebug_src_.png, WILDFLYHOME____user_projects____WFLY9542_postgres.7z, deploy.log, module_to_upload_to_red_hat.7z, server.log, wildlfly-ironjacamar-unregisterconnection.7z, wildlfly-ironjacamar-unregisterconnection.7z I have been trying to understand for what reason we sometimes are seing the exception: {panel} IJ000311: Throwable from unregister connection: java.lang.IllegalStateException: IJ000152: Trying to return an unknown connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@2021ff6. {panel} I have now created a sample application that allows the exception to be reproduced: https://github.com/99sono/wildlfly-ironjacamar-unregisterconnection/ To summarize: If we start a JTA transaction T1 by invoking an @EJB with TransacitonAttirbute RequiresNew. And then this EJB calls a second EJB with TransacitonAttirbute Supports. Then we will have the problem that if the First JDBC access is done on the nested EJB, the connection handed over to the JPA layer (eclipselink) will be popped out of the org.jboss.jca.core.connectionmanager.ccm.CachedConnectionManagerImpl.currentObjects As soon as we come out of the Nested EJB call. I do not know how serious this is. If somehow the popped out connection continues in fact locked to the ongoing JTA transaction or if might be going back to the connection pool and potentially be up for grabs. This I do not know how... I lack the knowledge to say how crucial it is to have a good synchrinization between the connections being managed under the: org.jboss.jca.core.connectionmanager.ccm.CachedConnectionManagerImpl And the JPA layer. But intuitively, I would say: quite important. So whenever a Stack gets popped with connections being managed, I would intuitively say: one of the parties has done something wrong. Either: (a) ARJUNA is asking JAC AMAR to open stacks when the JTA transaction is the same and it shouldn't, or (b) It does not really matter that the conneciton is popped out - somehow JAC AMAR still knows that the connection it popped out is still bound to the JTA transaction and the problem is logically strange but of low impact... What appears to be true, is that eclipselink is doing something logicla, in the sense that it is only asking for the conneciton to be closed/unregistered at the end of the JTA transaction. NOTE: In fact the error that we see when we get the stack trace, could also happen and no stack trance be logged by jac AMAR. This is the scenario when JAC leaves the unregister conneciton via: org.jboss.jca.core.connectionmanager.ccm.CachedConnectionManagerImpl would be going out via the: {panel} CopyOnWriteArrayList<ConnectionRecord> conns = cmToConnectionsMap.get(cm); // Can happen if connections are "passed" between contexts if (conns == null) return; {panel} But in my oppinion this exit path deserves the exact same stack trace. The sample application is to be found here: https://github.com/99sono/wildlfly-ironjacamar-unregisterconnection/ This sample application is not as simple to deploy because one needs to hava a data source configured. In any case the sample application has in src/test/resources my standalone.xml. It should be realtively easy to fix the JDBC drivers configured in the atnadlone.xml. And it should be relatively simple to fix the persistence.xml to search for eclipselink in the appropriate moduel. I will no copy paste bellow the LOG of the exception of the sample application. {panel} 2017-11-09 10:20:00,918 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) -------------------------------------- -- STARTING TransactionImple < ac, BasicAction: 0:ffffac140c58:634ef22a:5a041dbd:15 status: ActionStatus.RUNNING >: observing the CDI event StartupEventToTryToReproduceIJ000311Issue -------------------------------------- 2017-11-09 10:20:00,919 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) Doing a fetch ALL query. just to bind the connection manager to the KEY of stack level 1. 2017-11-09 10:20:01,033 FINE [org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor] (ServerService Thread Pool -- 73) Eclipselink connected (jndiName: java:/jdbc/SAMPLE_DS)- obtaining connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@1d5b3a43. Eclipselink Server session is: ServerSession( DatabaseAccessor(connected) PostgreSQLPlatform) 2017-11-09 10:20:01,057 FINE [org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor] (ServerService Thread Pool -- 73) EclipseLink - Closing connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@1d5b3a43 2017-11-09 10:20:01,063 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) At this point the org.jboss.jca.core.connectionmanager.ccm.CachedConnectionManagerImpl will have registered on stack1 the connection used by eclipselink, and unregistered as well. However, this temporary register/de-register is sufficient to add to the current key.getCMToConnectionsMap() and entry for the JDBC connection manger. This is fundamental, because without this step the exception that we will later see would be hidden away - even though the logical problem would be there no exception would be logged. By taking this magical step here, we have ensured that we will be able to reproduce the excpetion later. To summarize what has happend so far. 1. We have forced eclipselink to register and unregister a connection due to are fetch id query 2. The JTA transaction ID TransactionImple < ac, BasicAction: 0:ffffac140c58:634ef22a:5a041dbd:15 status: ActionStatus.RUNNING > continues healthy and running without any problem 3. Meanwhile internally, our CachedConnectionManagerImpl, for this thread, has a single task of keys in its currentObjects field 4. The leading key on this stack does not hold any connectionrecords (eclipselink returned the connecton to the pool) but it does have a KEY for the JDBC connection manager 5. Point 4 is the magical step to allow us to see the stack trace but not the core of the issue - just the core of seing the stack trace 2017-11-09 10:20:01,066 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) START: executoreSynchronousSameJta 2017-11-09 10:20:01,066 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) Right now we are within the same JTA transaction. But on a NESTED EJB call. What technically has happened at this point is that IRON JAC AMAR now has a second stack (keyAssociation) in the currentObjects. We add in stacks each time we go through an EJB evne if we keep the JTA transaction and this is the BUG. But we will see more about this later. 2017-11-09 10:20:01,066 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) STARTING 3: observing the CDI event StartupEventToTryToReproduceIJ000311Issue - The current transaction id is: TransactionImple < ac, BasicAction: 0:ffffac140c58:634ef22a:5a041dbd:15 status: ActionStatus.RUNNING > 2017-11-09 10:20:01,066 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) Going to create 50 entities on the database 2017-11-09 10:20:01,067 FINE [org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor] (ServerService Thread Pool -- 73) Eclipselink connected (jndiName: jdbc/SAMPLE_NON_JTA_DS)- obtaining connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@4be2573e. Eclipselink Server session is: ClientSession({}) 2017-11-09 10:20:01,073 FINE [org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor] (ServerService Thread Pool -- 73) EclipseLink - Closing connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@4be2573e 2017-11-09 10:20:01,078 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) 50 entities should have been created: 2017-11-09 10:20:01,078 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) Now we want to flush - we want to force eclipselink to have to acquire the java:/jdbc/SAMPLE_DS and not be allowed to close the connection yet. 2017-11-09 10:20:01,085 FINE [org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor] (ServerService Thread Pool -- 73) Eclipselink connected (jndiName: java:/jdbc/SAMPLE_DS)- obtaining connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@2021ff6. Eclipselink Server session is: ClientSession({default=DatabaseAccessor(connected)}) 2017-11-09 10:20:01,101 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) The flush has been done. We expec that the log now shows us that java:/jdbc/SAMPLE_DS connection was acquired but not yet closed by eclipselink. 2017-11-09 10:20:01,101 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) ENDING TransactionImple < ac, BasicAction: 0:ffffac140c58:634ef22a:5a041dbd:15 status: ActionStatus.RUNNING >: finishing to observer. The current transaction id is: {} 2017-11-09 10:20:01,101 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) OK. Now we are about to leave our nested EJB call. What is happening now is very important. 1. We will leave a nested EJB call but not the JTA transaction 2. Wildfly will go to the cachedConnectonManagerImpl and tell it to pop a stack (BUG) 3. The stack that goes away is actually holding the JDBC connection thta has been given to ECLIPSELINK and thta eclipselink will later try to close 4. When we leave this call we go back to STACK 1, where no connections are being managed 2017-11-09 10:20:01,102 INFO [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) FINISH: executoreSynchronousSameJta: 2017-11-09 10:20:01,102 WARN [startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb] (ServerService Thread Pool -- 73) NOW we are out of our nested EJB call. We are bout to leave the main EJB transaction call. In short we are about to reproduce the stack trace. What will now happen is the following. 1. Wildfly will tell eclipselink that the transaction is at an end and it should commit if it can 2. Eclipselink does its job of committing the changes and takes the JTA connection it was given and tries to close it 3. IRON JAC AMAR will log the stack trace we wanted to reproduce saying it has not idea what connection we are trying to return to the pool. 4. The reason the CachedConnectionManagerImpl does not know about the connection is because it was managed under STACK 2 but was popped out of existence after the nested transaction 5. Now it is very important to keep in mind that even if we were not seing the stack trace the error would be there (the moment we popped a key stack that had still connections) 6. Keep in mind that if we were to modify the source code and disable our reproduceBugService.fetchAlIds() call we would see no stack trace but the same issue would be present.7. The difference that doing the reproduceBugService.fetchAlIds() does is whether or not CachedConnectionManagerImpl during unregister connection leaves via the if conns == null return null or if goes further into the if (!ignoreConnections) 8. OK - Now that we understand what is happening behind the scenes we can let eclipselink try to close the JDBC connection at the end of the JTA transaction 2017-11-09 10:20:01,106 FINE [org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor] (ServerService Thread Pool -- 73) EclipseLink - Closing connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@2021ff6 2017-11-09 10:20:01,107 INFO [org.jboss.jca.core.connectionmanager.listener.TxConnectionListener] (ServerService Thread Pool -- 73) IJ000311: Throwable from unregister connection: java.lang.IllegalStateException: IJ000152: Trying to return an unknown connection: org.jboss.jca.adapters.jdbc.jdk7.WrappedConnectionJDK7@2021ff6 at org.jboss.jca.core.connectionmanager.ccm.CachedConnectionManagerImpl.unregisterConnection(CachedConnectionManagerImpl.java:408) at org.jboss.jca.core.connectionmanager.listener.TxConnectionListener.connectionClosed(TxConnectionListener.java:645) at org.jboss.jca.adapters.jdbc.BaseWrapperManagedConnection.returnHandle(BaseWrapperManagedConnection.java:596) at org.jboss.jca.adapters.jdbc.BaseWrapperManagedConnection.closeHandle(BaseWrapperManagedConnection.java:541) at org.jboss.jca.adapters.jdbc.WrappedConnection.returnConnection(WrappedConnection.java:298) at org.jboss.jca.adapters.jdbc.WrappedConnection.close(WrappedConnection.java:256) at org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor.closeDatasourceConnection(DatabaseAccessor.java:522) at org.eclipse.persistence.internal.databaseaccess.DatasourceAccessor.closeConnection(DatasourceAccessor.java:537) at org.eclipse.persistence.internal.databaseaccess.DatabaseAccessor.closeConnection(DatabaseAccessor.java:545) at org.eclipse.persistence.internal.databaseaccess.DatasourceAccessor.closeJTSConnection(DatasourceAccessor.java:188) at org.eclipse.persistence.sessions.server.ClientSession.releaseJTSConnection(ClientSession.java:171) at org.eclipse.persistence.transaction.AbstractSynchronizationListener.beforeCompletion(AbstractSynchronizationListener.java:175) at org.eclipse.persistence.transaction.JTASynchronizationListener.beforeCompletion(JTASynchronizationListener.java:68) at org.jboss.as.txn.service.internal.tsr.JCAOrderedLastSynchronizationList.beforeCompletion(JCAOrderedLastSynchronizationList.java:116) at com.arjuna.ats.internal.jta.resources.arjunacore.SynchronizationImple.beforeCompletion(SynchronizationImple.java:76) at com.arjuna.ats.arjuna.coordinator.TwoPhaseCoordinator.beforeCompletion(TwoPhaseCoordinator.java:368) at com.arjuna.ats.arjuna.coordinator.TwoPhaseCoordinator.end(TwoPhaseCoordinator.java:91) at com.arjuna.ats.arjuna.AtomicAction.commit(AtomicAction.java:162) at com.arjuna.ats.internal.jta.transaction.arjunacore.TransactionImple.commitAndDisassociate(TransactionImple.java:1200) at com.arjuna.ats.internal.jta.transaction.arjunacore.BaseTransaction.commit(BaseTransaction.java:126) at com.arjuna.ats.jbossatx.BaseTransactionManagerDelegate.commit(BaseTransactionManagerDelegate.java:89) at org.jboss.as.ejb3.tx.CMTTxInterceptor.endTransaction(CMTTxInterceptor.java:91) at org.jboss.as.ejb3.tx.CMTTxInterceptor.invokeInOurTx(CMTTxInterceptor.java:279) at org.jboss.as.ejb3.tx.CMTTxInterceptor.requiresNew(CMTTxInterceptor.java:349) at org.jboss.as.ejb3.tx.CMTTxInterceptor.processInvocation(CMTTxInterceptor.java:241) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.interceptors.CurrentInvocationContextInterceptor.processInvocation(CurrentInvocationContextInterceptor.java:41) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.invocationmetrics.WaitTimeInterceptor.processInvocation(WaitTimeInterceptor.java:47) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.security.SecurityContextInterceptor.processInvocation(SecurityContextInterceptor.java:100) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.deployment.processors.StartupAwaitInterceptor.processInvocation(StartupAwaitInterceptor.java:22) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.InterceptorContext$Invocation.proceed(InterceptorContext.java:437) at org.jboss.as.ejb3.concurrency.ContainerManagedConcurrencyInterceptor.processInvocation(ContainerManagedConcurrencyInterceptor.java:110) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.interceptors.ShutDownInterceptorFactory$1.processInvocation(ShutDownInterceptorFactory.java:64) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.interceptors.LoggingInterceptor.processInvocation(LoggingInterceptor.java:67) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ee.component.NamespaceContextInterceptor.processInvocation(NamespaceContextInterceptor.java:50) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.interceptors.AdditionalSetupInterceptor.processInvocation(AdditionalSetupInterceptor.java:54) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.ContextClassLoaderInterceptor.processInvocation(ContextClassLoaderInterceptor.java:64) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.InterceptorContext.run(InterceptorContext.java:356) at org.wildfly.security.manager.WildFlySecurityManager.doChecked(WildFlySecurityManager.java:636) at org.jboss.invocation.AccessCheckingInterceptor.processInvocation(AccessCheckingInterceptor.java:61) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.InterceptorContext.run(InterceptorContext.java:356) at org.jboss.invocation.PrivilegedWithCombinerInterceptor.processInvocation(PrivilegedWithCombinerInterceptor.java:80) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.ChainedInterceptor.processInvocation(ChainedInterceptor.java:61) at org.jboss.as.ee.component.ViewService$View.invoke(ViewService.java:198) at org.jboss.as.ee.component.ViewDescription$1.processInvocation(ViewDescription.java:185) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.ChainedInterceptor.processInvocation(ChainedInterceptor.java:61) at org.jboss.as.ee.component.ProxyInvocationHandler.invoke(ProxyInvocationHandler.java:73) at startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb$$$view9.observeStartupEventToTryToReproduceIJ000311Issue(Unknown Source) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:498) at org.jboss.weld.util.reflection.Reflections.invokeAndUnwrap(Reflections.java:433) at org.jboss.weld.bean.proxy.EnterpriseBeanProxyMethodHandler.invoke(EnterpriseBeanProxyMethodHandler.java:128) at org.jboss.weld.bean.proxy.EnterpriseTargetBeanInstance.invoke(EnterpriseTargetBeanInstance.java:56) at org.jboss.weld.bean.proxy.ProxyMethodHandler.invoke(ProxyMethodHandler.java:100) at startupconnectionproblem.observer.EventObserverNestedJtaTransactionEjb$Proxy$_$$_Weld$EnterpriseProxy$.observeStartupEventToTryToReproduceIJ000311Issue(Unknown Source) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:498) at org.jboss.weld.injection.StaticMethodInjectionPoint.invoke(StaticMethodInjectionPoint.java:88) at org.jboss.weld.injection.StaticMethodInjectionPoint.invoke(StaticMethodInjectionPoint.java:78) at org.jboss.weld.injection.MethodInvocationStrategy$SimpleMethodInvocationStrategy.invoke(MethodInvocationStrategy.java:129) at org.jboss.weld.event.ObserverMethodImpl.sendEvent(ObserverMethodImpl.java:313) at org.jboss.weld.event.ObserverMethodImpl.sendEvent(ObserverMethodImpl.java:291) at org.jboss.weld.event.ObserverMethodImpl.notify(ObserverMethodImpl.java:269) at org.jboss.weld.event.ObserverNotifier.notifySyncObservers(ObserverNotifier.java:302) at org.jboss.weld.event.ObserverNotifier.notify(ObserverNotifier.java:291) at org.jboss.weld.event.EventImpl.fire(EventImpl.java:89) at startupconnectionproblem.event.FireStartupEventsServiceImpl.fireStartupCDIEvents(FireStartupEventsServiceImpl.java:41) at startupconnectionproblem.event.FireStartupEventsServiceImpl$Proxy$_$$_WeldClientProxy.fireStartupCDIEvents(Unknown Source) at startupconnectionproblem.event.StartupSingletonLocalEjb.postConstrcut(StartupSingletonLocalEjb.java:38) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:498) at org.jboss.as.ee.component.ManagedReferenceLifecycleMethodInterceptor.processInvocation(ManagedReferenceLifecycleMethodInterceptor.java:96) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.weld.ejb.Jsr299BindingsInterceptor.doLifecycleInterception(Jsr299BindingsInterceptor.java:114) at org.jboss.as.weld.ejb.Jsr299BindingsInterceptor.processInvocation(Jsr299BindingsInterceptor.java:103) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.InterceptorContext$Invocation.proceed(InterceptorContext.java:437) at org.jboss.weld.ejb.AbstractEJBRequestScopeActivationInterceptor.aroundInvoke(AbstractEJBRequestScopeActivationInterceptor.java:73) at org.jboss.as.weld.ejb.EjbRequestScopeActivationInterceptor.processInvocation(EjbRequestScopeActivationInterceptor.java:83) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.weld.injection.WeldInjectionInterceptor.processInvocation(WeldInjectionInterceptor.java:53) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ee.component.AroundConstructInterceptorFactory$1.processInvocation(AroundConstructInterceptorFactory.java:28) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.weld.injection.WeldInterceptorInjectionInterceptor.processInvocation(WeldInterceptorInjectionInterceptor.java:56) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.weld.ejb.Jsr299BindingsCreateInterceptor.processInvocation(Jsr299BindingsCreateInterceptor.java:100) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ee.component.NamespaceContextInterceptor.processInvocation(NamespaceContextInterceptor.java:50) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.tx.CMTTxInterceptor.invokeInNoTx(CMTTxInterceptor.java:263) at org.jboss.as.ejb3.tx.LifecycleCMTTxInterceptor.notSupported(LifecycleCMTTxInterceptor.java:87) at org.jboss.as.ejb3.tx.LifecycleCMTTxInterceptor.processInvocation(LifecycleCMTTxInterceptor.java:64) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.weld.injection.WeldInjectionContextInterceptor.processInvocation(WeldInjectionContextInterceptor.java:43) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.interceptors.CurrentInvocationContextInterceptor.processInvocation(CurrentInvocationContextInterceptor.java:41) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ee.concurrent.ConcurrentContextInterceptor.processInvocation(ConcurrentContextInterceptor.java:45) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.ContextClassLoaderInterceptor.processInvocation(ContextClassLoaderInterceptor.java:64) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.as.ejb3.component.singleton.StartupCountDownInterceptor.processInvocation(StartupCountDownInterceptor.java:25) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.InterceptorContext.run(InterceptorContext.java:356) at org.jboss.invocation.PrivilegedWithCombinerInterceptor.processInvocation(PrivilegedWithCombinerInterceptor.java:80) at org.jboss.invocation.InterceptorContext.proceed(InterceptorContext.java:340) at org.jboss.invocation.ChainedInterceptor.processInvocation(ChainedInterceptor.java:61) at org.jboss.as.ee.component.BasicComponent.constructComponentInstance(BasicComponent.java:161) at org.jboss.as.ee.component.BasicComponent.constructComponentInstance(BasicComponent.java:134) at org.jboss.as.ee.component.BasicComponent.createInstance(BasicComponent.java:88) at org.jboss.as.ejb3.component.singleton.SingletonComponent.getComponentInstance(SingletonComponent.java:124) at org.jboss.as.ejb3.component.singleton.SingletonComponent.start(SingletonComponent.java:138) at org.jboss.as.ee.component.ComponentStartService$1.run(ComponentStartService.java:54) at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) at java.util.concurrent.FutureTask.run(FutureTask.java:266) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748) at org.jboss.threads.JBossThread.run(JBossThread.java:320) 2017-11-09 10:20:01,114 INFO [startupconnectionproblem.event.FireStartupEventsServiceImpl] (ServerService Thread Pool -- 73) FIINSH: StartupEventToTryToReproduceIJ000311Issue 2017-11-09 10:20:01,114 INFO [startupconnectionproblem.event.StartupSingletonLocalEjb] (ServerService Thread Pool -- 73) ENDING 1: Post construct of startup bean singleton ending 2017-11-09 10:20:01,240 INFO [javax.enterprise.resource.webcontainer.jsf.config] (ServerService Thread Pool -- 73) Initializing Mojarra 2.2.13.SP1 20160303-1204 for context '/app' 2017-11-09 10:20:02,098 INFO [org.primefaces.webapp.PostConstructA {panel} Thanks for considering this issue. I do not know how much impact this issue can actually have. We have many hundred different types of JTA business transactions, and we see this sample application happening on system tests and stress tests relatively often. Now that I have better understand of the issue, I can in some cases rework the code to dodge this problem to avoid nested calls of EJBS on the same transaciton - but this will not always be possible. And I would be quite afraid of this issue, if means that our connectons still in the hands of ElcipseLInk are going back to the pool. This would explain some strange errors we have had only on stress test application, that we do not see when using a different app server. I must say it was extremelly difficult to be able to make the stack trace reproducible. There are many tricky points to be able to lead the code to go precisely through the code path that leads to the excpetion. But it is relatively trivial to make the code leave through the "silent bug iron jac amar code path". Many thanks for all the help on this.