Red Hat Bugzilla – Bug 1304325
Dead lock between Topology updates and connection creation
Last modified: 2016-06-14 07:37:48 EDT
Created attachment 1120714 [details]
log produced by org.hornetq.tests.integration.jms.cluster.MultipleThreadsOpeningTest.testMultipleOpen
Description of problem:
In a rare case deadlock between topology updates and connection creation may occur. This issue was fixed for ActiveMQ Artemis, but not for HornetQ. Complete information and description of this issue can be found here https://issues.apache.org/jira/browse/ARTEMIS-217
Attaching log of test org.hornetq.tests.integration.jms.cluster.MultipleThreadsOpeningTest.testMultipleOpen where this issue can be hit.
Fix in ARTEMIS-217 caused another issue described in: https://issues.jboss.org/browse/JBEAP-1713
Fix from ARTEMIS-217 should be adjusted not to block update of "receivedTopology" attribute.
not really, (unless there's another lock in place, causing a deadlock)... which I don't see indications here.
this is using wait/notify pattern, which means the lock will be released while the wait is in place.
I meant to say.. I don't think the issue is caused by the change on ARTEMIS-217. If there's an issue it's something else.
I can't replicate it BTW?
I got confused, thought this was a new request.. fixing it as part of JBEAP-1713
I fixed JBEAP-1713 but I don't see any relation to this previously fixed issue on ARTEMIS-217.
I've sent a fix as part of JBEAP-1713: https://github.com/apache/activemq-artemis/pull/369 which will be back ported into the JBoss EAP 7 branch
ARTEMIS-217 added a synchronized block which delayed/blocked update of receivedTopology attribute in createSessionFactory() method and throw exception. I did not have time to study it deeply so it could be something else as well.
MultipleThreadsOpeningTest.testMultipleOpen was failing on Artemis bacause of JBEAP-1713.
As we did not see fix ARTEMIS-217 to be backported to HornetQ and saw that MultipleThreadsOpeningTest.testMultipleOpen test fails sometimes due to client not finished. It indicated that dead lock might be there similar to ARTEMIS-217 could be still in HornetQ.
Erich Duda <email@example.com> updated the status of jira JBEAP-1713 to Resolved
Jiri Pallich <firstname.lastname@example.org> updated the status of jira JBEAP-1713 to Closed