Hi,
Our team is pretty new to JBI and OpenESB, so we're still trying to get our heads around it and gain some level of comfort. Actually we are using JCAPS 6, and deploying to Enterprise Glassfish 2.1. I'm not exactly sure whether this is a question for this group, or for OpenESB, so I'm cross posting to both groups.
The situation we have encountered is when doing a redeploy of a Service Assembly to a cluster in Glassfish. We can see that the new version was successfully deployed to the jbi/service-assemblies directory of the domain (DAS), but the instances (under the node agent) still have the previous version.
All deployments are handled by our middleware team, for whom JBI is just as new a concept as it is for us. Because they have encountered issues in the past deploying J2EE apps while the cluster was running, their procedure is to shut down the cluster, undeploy the existing application (if there is one), deploy the new application, and restart the cluster. Upon restart of the cluster instances the node agent syncs the cluster instances up with the DAS so that they have the new application.
It seems that this procedure is not valid for a Service Assembly? As a test, the middleware team performed a redeploy without shutting down the cluster, and the Service Assembly was then successfully deployed out to the cluster instances.
So I'm really looking for confirmation that when deploying a Service Assembly it is necessary to have all instances to which it is to be deployed running. Is this also the case for other JBI components (shared libraries and binding components)? Also, is this by design, or is it a bug?
Thanks,
Dave Derry
[Message sent by forum member 'davederry' (dave.derry_at_equifax.com)]
http://forums.java.net/jive/thread.jspa?messageID=362809