[jboss-jira] [JBoss JIRA] (WFWIP-87) [Artemis 2.x upgrade] Transaction remained in prepared state after failover of MDB

[Clebert Suconic (JIRA)]

    [ https://issues.jboss.org/browse/WFWIP-87?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13620313#comment-13620313 ] 

Clebert Suconic commented on WFWIP-87:
--------------------------------------

[~mnovak]: 

I could find what's the issue causing this. The testcase I wrote even though it can't be merged on master yet could already tell me what's going on.


The replicationManager is clearing up tokens on replication. however there's some code in 1.x to clear these tokens before shutdown. what is causing the issue between the two branches.


Is there any chance you can try this branch on the test?

https://github.com/clebertsuconic/activemq-artemis/tree/replication-test


> [Artemis 2.x upgrade] Transaction remained in prepared state after failover of MDB
> ----------------------------------------------------------------------------------
>
>                 Key: WFWIP-87
>                 URL: https://issues.jboss.org/browse/WFWIP-87
>             Project: WildFly WIP
>          Issue Type: Bug
>          Components: Artemis
>            Reporter: Miroslav Novak
>            Assignee: Justin Bertram
>            Priority: Blocker
>
> There is regression in test scenario where MDB failovers from live to backup. Live is cleanly shutdown. 
> Problem is that after MDB failovers to backup then there is still transaction in prepared state which survive periodic recovery cycles. 
> Link to build:
> https://eap-qe-jenkins.rhev-ci-vms.eng.rdu2.redhat.com/job/eap-7.x-messaging-weekly-ha-failover-dedicated-mdb-replicated-journal/26/testReport/junit/org.jboss.qa.hornetq.test.failover/ReplicatedDedicatedFailoverTestWithMdb/testShutdown/



--
This message was sent by Atlassian JIRA
(v7.5.0#75005)