I am using JBoss 5.1 (Enterprise Edition) to deply a war-file to a JBoss 5.1 cluster. In testing deployment features of JBoss, we always deployed a war-file to a JBoss 5.1 cluster by putting it into the farm/ directory of one of the instances of the cluster. It works perfectly because the hdscanner is running.
Going to production with the cluster, we probably don't want the hdscanner anymore. My question is: are there ways to deploying to a clustetr without a hdscanner running. I got some answers to my post http://community.jboss.org/message/585115#585115Deployment with twiddle / twiddle documentation whle I was looking into twiddle. While twiddle is great in retrieving all kinds of useful information, it deploys the application directly to a tmp/ directory (which has as main drawback that this must be done every time JBoss starts). I also haven't been able to find an MBean that deploys to a cluster. Another drawback is that twiddle deploys to tmp/<UUID>/application.war/, where <UUID> is something like 41263dx-69rwqb-gjply6fw-1-gjqrbq6z-aw, every time the server is restarted, a new UUID is generated, thus generating clutter on the server.
Question is: Is there a way to deploy a war file (or any file, for that matter) to a clustered environment where hdscanner is not used (something in the line of: put the to-deployed files it in the farm/ directory of one of the nodes, and signal JBoss to look for once in the farm/ directory for "pending" deployments).? I like to know if such a signal exist and how to call it, in this case.