I now have things working with the default JBoss 3.0.6 distribution, the only modification being the jboss_init_redhat.sh script, where I replace run.sh with run.sh -c all.
However, as soon as I deploy to the farm directory, I still run into the memory issues. I see no memory discussion in the clustering pdf. How big are the systems these clusters are being deployed on? Our production servers are bigger than our workstations, but I don't want to obliterate a production server the way I'm trashing my workstations and test boxes.