[jboss-jira] [JBoss JIRA] (JBMESSAGING-1931) Message not delivered to Topic afer node rejions cluster

[Yong Hao Gao (JIRA)]

    [ https://issues.jboss.org/browse/JBMESSAGING-1931?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12709060#comment-12709060 ] 

Yong Hao Gao commented on JBMESSAGING-1931:
-------------------------------------------

Thanks Doug. It seems that when you disconnect node 2 in the test, the DB is also disconnected from node 1. Node 1 is watching the timestamp of node 2 by periodically querying the JBM_CLUSTER table in the DB. If node 1 found it cannot see the timestamp of node 2 be updated for sometime (basically it is about two times the value of NodeStateRefreshInterval, i.e. 2*20 = 40 sec in this test), it will think that node 2 is really dead and perform failover for it. In the test you waited for 1 min, so the node 1 (as shown in the log) eventually performed failover for node 2 before you reconnected it, and the cluster will not be re-formed correctly.

I'd suggest you increase NodeStateRefreshInterval to more than 30 sec to make sure failover won't happen before your waiting time, and redo the test and see how it goes.

Howard

                
> Message not delivered to Topic afer node rejions cluster
> --------------------------------------------------------
>
>                 Key: JBMESSAGING-1931
>                 URL: https://issues.jboss.org/browse/JBMESSAGING-1931
>             Project: JBoss Messaging
>          Issue Type: Bug
>          Components: JMS Clustering
>    Affects Versions: 1.4.8.SP5
>         Environment: JBoss EAP 5.1.2
>            Reporter: Doug Grove
>            Assignee: Yong Hao Gao
>         Attachments: node1-config.tar.gz, node1_logs.tar.gz, node2-config.tar.gz, node2_logs.tar.gz
>
>
> After a node leaves the cluster and then rejoins the cluster, the node does not receive messages posted to clustered topics that are delivered to other nodes.
> Log files are attached.  Node 2 was the node taken out of the cluster.  After Node 2 was allowed to rejoin the cluster, a message was sent to Node 1.  It was not seen by Node 2.  A messages was then sent to Node 2.  It was seen by Node 1.  Search for "Hello World" in the server.log.
> Note that this configuration has the cluster health mbean installed.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.jboss.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira