Managing JBoss Cluster

[ivelin.ivanov]

The latest JBoss book says that nodes participating in a cluster cannot be currently monitored nor managed via centralized console.
WL7 has the feature
http://edocs.bea.com/wls/docs70/jmx/runtime.html#1122173

Is this something on the current roadmap. If I was interested to help move this forward, where would I start?

[adrian.brock]

JBoss head sort of has this feature.
It uses cascading agents so that each agent
can see all mbeans across the network.
It is not currently integrated with the clustering
framework however, it requires separate
configuration.

To introduce this feature, we need use cases.
e.g. Should a change on machine replicate
across the cluster? How do you define this?
Which configuration should come from the cluster
when a machine joins rather than the local config.

This would be better posted in the jmx development
forum further down. You might want to post
a pointer to it in the clustering forum.

Regards,
Adrian

[ivelin.ivanov]

JSR 77 answers some of the use case question.
It addresses the issue of running a j2ee server and its components across multiple VMs.
Apparently there is already initiative to support JSR 77 within JBoss, but I cannot tell how much is clustering addressed.

[tomerbd]

lets say I have 20 nodes in my cluster, and i want to update one of my properties in my managed beans, that means i have to go (i use the jmx-console) through each node and configure it! its gonig to be a nightmare right? or are there other ways?

[dimitris]

Easiest thing to do is to deploy your service in the ./farm folder, so that re-deployment with new settings is propagates to all nodes.

However, you cannot use the jmx-console for this. This is a feature that will be probably supported by the new admin console.