[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño updated ISPN-1056: ----------------------------------- Fix Version/s: 4.2.2.BETA1 5.0.0.CR1 5.0.0.FINAL -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-1056: ---------------------------------------- Right, it seems like TransactionTable.stop() and CacheManagerNotifierImpl.stop() have the same stop priorities so sometimes the notifier might stop before transaction table. However, clearly TransactionTable.stop() needs to happen before CacheManagerNotifierImpl.stop() in order to make sure the listener that checks for stale transactions is properly unregistered. So, shifting stop priorities should work. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-1056: ---------------------------------------- Paolo, what you're seeing does not make sense at first glance. Could you please enable TRACE logging on org.infinispan package, repeat your test and upload the TRACE log? Thanks -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] paolo di tommaso commented on ISPN-1056: ---------------------------------------- Quite strange, I'm unable to reproduce it. I've set Infinispan logging to TRACE, but now I'm getting the following trace on stopping the server: 38160 [Thread-2] TRACE org.infinispan.marshall.jboss.ConstantObjectTable - Externalizer reader and writer maps have been cleared and constant object table was stopped 38161 [Thread-2] DEBUG org.infinispan.lifecycle.ComponentStatus - Ignoring call to stop() as current state is STOPPING 38161 [Thread-2] TRACE org.infinispan.factories.GlobalComponentRegistry - Reset volatile components. Registry now contains [org.infinispan.config.GlobalConfiguration, org.infinispan.jmx.CacheManagerJmxRegistration, java.lang.ClassLoader, org.infinispan.manager.EmbeddedCacheManager, org.infinispan.factories.GlobalComponentRegistry] 38164 [Thread-4] INFO play - >>> Stopping server 38165 [Thread-5] TRACE org.infinispan.marshall.jboss.ConstantObjectTable - Externalizer reader and writer maps have been cleared and constant object table was stopped 38167 [Thread-5] DEBUG org.infinispan.lifecycle.ComponentStatus - Ignoring call to stop() as current state is STOPPING 38168 [Thread-5] TRACE org.infinispan.factories.GlobalComponentRegistry - Reset volatile components. Registry now contains [org.infinispan.config.GlobalConfiguration, org.infinispan.jmx.CacheManagerJmxRegistration, java.lang.ClassLoader, org.infinispan.manager.EmbeddedCacheManager, org.infinispan.factories.GlobalComponentRegistry] 38172 [Thread-4] WARN play - Unable to invoke method private void org.infinispan.transaction.xa.TransactionTable.stop() on object -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-1056: ---------------------------------------- Paolo, can you please post the entire file with the TRACE log on? There appears to be a mix of threads there and so I suspect that this is an issue with concurrent calls to either CacheManager.stop() and/or Cache.stop(). In fact, what's probably happening is that both Thread-2 and Thread-5 are stopping the CacheManager (ConstantObjectTable is a variable of the marshaller which is a global component) and Thread-4 is stopping one of the caches. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] paolo di tommaso commented on ISPN-1056: ---------------------------------------- OK, finally I got the issue stack trace on application stop. [Thread-5] TRACE org.infinispan.marshall.jboss.ConstantObjectTable - Externalizer reader and writer maps have been cleared and constant object table was stopped 32455 [Thread-5] DEBUG org.infinispan.lifecycle.ComponentStatus - Ignoring call to stop() as current state is STOPPING 32455 [Thread-5] TRACE org.infinispan.factories.GlobalComponentRegistry - Reset volatile components. Registry now contains [org.infinispan.config.GlobalConfiguration, org.infinispan.jmx.CacheManagerJmxRegistration, java.lang.ClassLoader, org.infinispan.manager.EmbeddedCacheManager, org.infinispan.factories.GlobalComponentRegistry] 32456 [Thread-3] TRACE org.infinispan.marshall.jboss.ConstantObjectTable - Externalizer reader and writer maps have been cleared and constant object table was stopped 32456 [Thread-3] DEBUG org.infinispan.lifecycle.ComponentStatus - Ignoring call to stop() as current state is STOPPING 32457 [Thread-3] TRACE org.infinispan.factories.GlobalComponentRegistry - Reset volatile components. Registry now contains [org.infinispan.config.GlobalConfiguration, org.infinispan.jmx.CacheManagerJmxRegistration, java.lang.ClassLoader, org.infinispan.manager.EmbeddedCacheManager, org.infinispan.factories.GlobalComponentRegistry] 32456 [Thread-4] INFO play - >>> Stopping server Exception in thread "Thread-4" org.infinispan.CacheException: Unable to invoke method private void org.infinispan.transaction.xa.TransactionTable.stop() on object at org.infinispan.util.ReflectionUtil.invokeAccessibly(ReflectionUtil.java:174) at org.infinispan.factories.AbstractComponentRegistry$PrioritizedMethod.invoke(AbstractComponentRegistry.java:889) at org.infinispan.factories.AbstractComponentRegistry.internalStop(AbstractComponentRegistry.java:716) at org.infinispan.factories.AbstractComponentRegistry.stop(AbstractComponentRegistry.java:610) at org.infinispan.factories.ComponentRegistry.stop(ComponentRegistry.java:172) at org.infinispan.CacheDelegate.stop(CacheDelegate.java:331) at cache.InfinispanCacheImpl.stop(InfinispanCacheImpl.java:172) at play.cache.Cache.stop(Cache.java:247) at play.Play.stop(Play.java:503) at play.Play$1.run(Play.java:410) Caused by: java.lang.reflect.InvocationTargetException at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25) at java.lang.reflect.Method.invoke(Method.java:597) at org.infinispan.util.ReflectionUtil.invokeAccessibly(ReflectionUtil.java:171) ... 9 more Caused by: java.lang.NullPointerException at org.infinispan.manager.DefaultCacheManager.removeListener(DefaultCacheManager.java:567) at org.infinispan.transaction.xa.TransactionTable.stop(TransactionTable.java:110) ... 14 more -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-1056: ---------------------------------------- Maybe this can be resolved by making sure that the CacheManager does not stop the global component registry until all caches have reached the STOPPED stage. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] paolo di tommaso commented on ISPN-1056: ---------------------------------------- OK, I've just attached to the issue the trace file. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-1056: ---------------------------------------- Thanks Paolo. Seems like there really is not anything else in the file. Anyway, apart from my suggestion earlier, this can possibly be fixed in a much easier way by having a boolean volatile that is set to true when CacheManager.stop() is called, and if another comes through, instead of bypassing potentially stopping caches and stop the global component registry, it would simply ignore the call to CacheManager.stop(). That way, only the initial thread that called CacheManager.stop() will stop the global component registry once all caches are stopped which should stop the NPE. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Manik Surtani updated ISPN-1056: -------------------------------- Fix Version/s: 5.0.0.CR2 -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1056?page=com.atlassian.jira.plugin.... ] Galder Zamarreño updated ISPN-1056: ----------------------------------- Fix Version/s: 4.2.2.BETA1 -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira