[esb-issues] [JBoss JIRA] Commented: (JBESB-1456) Add documentation to describe the effect of clustered JMS queues on service invocations

[Martin Vecera (JIRA)]

    [ http://jira.jboss.com/jira/browse/JBESB-1456?page=comments#action_12394518 ] 
            
Martin Vecera commented on JBESB-1456:
--------------------------------------

Just an idea:

Warning: The load balancing happens every time the messages goes over clustered JMS queues. This means if you had a clustered service and the message was delivered through two clustered queues (usual case - the GW and ESB listener), the load balancing happens between the queues and can result to not delivering the messages to the service on some node, especially when bot queues are being served by more threads.

> Add documentation to describe the effect of clustered JMS queues on service invocations
> ---------------------------------------------------------------------------------------
>
>                 Key: JBESB-1456
>                 URL: http://jira.jboss.com/jira/browse/JBESB-1456
>             Project: JBoss ESB
>          Issue Type: Task
>      Security Level: Public(Everyone can see) 
>          Components: Documentation
>    Affects Versions: 4.2.1
>         Environment: SOA-P 4.2.0 beta 1
>            Reporter: Martin Vecera
>            Priority: Minor
>
> During failover testing I found out very interresting behaviour...
> I had two nodes (JBossESB servers) in a cluster.
> I deployed identical service to both of them. Each service had:
> 1) clustered gateway queue served by 2 threads
> 2) clustered queue for ESB aware messages served by 2 threads.
> I send several messages to the first node and expected them to be load balanced among the nodes. This didn't happend.
> I figured out that the load balancing came in to the play twice:
> 1) while sending the message to GW
> 2) while sending the message from GW to ESB-aware queue
> Because there were more than one thread for each queue the loadbalancing went in the way where some messages were really sent to the GW on the second node but then they returned back to the first one.
> This is definitely configuration failure (the queue for ESB aware messages mustn't be clustered) but we should document this I think.

-- 
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