Dynamic nodes addition thru apache plugin..

[ehskaga]

While configuring the apache plugin for jboss, it was observed that the list of app server nodes need to be statically defined in the worker.properties file. Does this infer that the list of app server is static for the plugin to load balance, so if another server (node) is dynamically added to the cluster (by auto-discovery feature) it will not be used by the plug-in.
This is being compared to the weblogic plug-in wherein the information about the available nodes are dynamically accounted for so that if a node is added or dropped, it is reflected correctly in the plug-in load balancer. While configuring it takes a static list of nodes to contact but on every invocation it downloads the cluster configuration w.r.t the nodes alive/dead and accrodingly changes its routing.

[brian.stansberry]

Yes, the worker list is statically defined. However, if you also use a status worker, apache will periodically (default once per minute) check the workers.properties file and load any updates. So, when you deploy a new node you can update workers.properties.

[ehskaga]

As i understand after reading through the status worker documentation, a capability exists to change the worker.properties manually and reload the configuration without starting the server using the status worker.
But this is totally against the philosphy of Auto-discovery feature available in JBoss, i mean if the application server instances can automatically discover nodes and dynamically converge as part of single cluster, the same should be achievable through the plug-in instead of a manual intervention requirement.

Kindly correct me if i am wrong.