[gatein-issues] [JBoss JIRA] (GTNWSRP-258) WSRP producer as 2-node cluster with loadbalancer - registration issue

[RH Bugzilla Integration (JIRA)]

    [ https://issues.jboss.org/browse/GTNWSRP-258?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12739264#comment-12739264 ] 

RH Bugzilla Integration commented on GTNWSRP-258:
-------------------------------------------------

claprun at redhat.com changed the Status of [bug 845535|https://bugzilla.redhat.com/show_bug.cgi?id=845535] from NEW to MODIFIED
                
> WSRP producer as 2-node cluster with loadbalancer - registration issue
> ----------------------------------------------------------------------
>
>                 Key: GTNWSRP-258
>                 URL: https://issues.jboss.org/browse/GTNWSRP-258
>             Project: GateIn WSRP
>          Issue Type: Sub-task
>      Security Level: Public(Everyone can see) 
>          Components: Producer
>    Affects Versions: 2.1.0-Beta08
>         Environment: EPP5.2.0 ER05
>            Reporter: Michal Vanco
>            Assignee: Chris Laprun
>             Fix For: 2.1.8-GA
>
>
> Imagine/configure the environment: producer as 2-node cluster with loadbalancer + 1 consumer.
> You register new consumer at consumer node with WSDL from loadbalancer. Loadbalancer points to one of nodes (let's wait node1) and registration is done with node1.
> btw. WSDL on loadbalancer directly points to one of nodes (there is node1 or node2 at all URLs)
> You put some remote portlet on page (communication is between consumer+node1 through loadbalancer), do failover - stop node1.
> At this moment you would expect that remote portlet is still available, but it's not because consumer is registered with node1.
> It means that after refresh of page with remote portlet -> portlet is not displayed.
> You have to go to WSRP admin and refresh consumer (with WSDL of loadbalancer), now the question:
>  - what is expected at this moment? Registration on second node? (this should be somehow distributed from killed node)
>  - when you have 2 cluster nodes started on one machine - you can't continue and refresh won't be successfull (Could not transmit message, Can not connect http client invoker after 1 attempt(s)) - this should be related to unique consumer name registration on producer
>  - when you have 2 nodes started on different machines - you should be able to refresh, what happens with registration? Is there new registration with second node? What happens when node1 is back? (2 different registrations for same consumer?)
> At this moment I've tested 2 nodes at same machine - at this scenario you can't continue using remote portlet and refresh consumer after failover.
> I can provide some more details based on reply of expected behavior.
> Thanks!

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira