JBossDeveloper

 

"scott.stark@jboss.org" wrote:
This has to be a side-effect of farming. Instead of farming the deployments, just deployment them to the servers and rerun the test. Farming has to be propagating the undeployment event to all nodes. This is not what the ProxyFactoryHA state listener was designed for.

I have now disabled farming on all nodes. I use the original ProxyFactoryHA again, and deploy the MBean manually. The odd behavior is still there. As soon as I undeploy the MBean on one node (delete .sar file from deploy directory) the MBean.STOPPING event is propagated to all replicants --> the containerIsAboutToStop() method gets called on every replicant --> they all disappear from the DistributedReplicantManager's listContent() view.. so my guess is that it has nothing to do with farming.

I would love to investigate further on this, as this has really sparked my interest, but unfortunately I have already spent more time on the MBean clustering than I was supposed to.. and the patch I posted above works fine for me.. so I think I'll just open a JIRA issue with the detailed behavior as soon as I find the time.

Thank you very much for your replies everybody. It's good to see that the devs are active in the forums and actually give valuable feedback to user questions. :-)