[jboss-jira] [JBoss JIRA] Created: (JBMESSAGING-667) Clustered ConnectionFactory binding mechanism broken for a single-node cluster (and possibly for multi-node cluster)

[Ovidiu Feodorov (JIRA)]

Clustered ConnectionFactory binding mechanism broken for a single-node cluster (and possibly for multi-node cluster)
--------------------------------------------------------------------------------------------------------------------

                 Key: JBMESSAGING-667
                 URL: http://jira.jboss.com/jira/browse/JBMESSAGING-667
             Project: JBoss Messaging
          Issue Type: Bug
          Components: JMS Clustering
    Affects Versions: 1.2.0.Beta1
            Reporter: Ovidiu Feodorov
         Assigned To: Ovidiu Feodorov
             Fix For: 1.2.0.Beta1


The current code that binds clustered ConnectionFactories in JNDI couldn't possibly run. Reasons:

* ConnectionFactoryJNDIMapper creates a "local" ClientConnectionFactoryDelegate, and then replicates it around the cluster. Because ConnectionFactoryJNDIMapper is a listener to its own replicant instance, it gets its own replication request and tries to add the ClientConnectionFactoryDelegate to a ClusteredConnectionFactoryDelegate that is not yet bound in the "factories" map. Hence, IllegalStateException("Cannot find connection factory " + uniqueName + " to update").
* Mixup between ClientConnectionFactoryDelegates and JBossConnectionFactories while adding to internal "factories" map.
* ConnectionFactoryJNDIMapper.setFailoverDelegates() indexes are messed up (the i index used instead of j in for (int j = 0; j < s; j++))
* etc. 

Before commenting on the above assertions, try running the simplest JMSTest in clustered mode.


-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira