The JBossClustering doc (JBossClustering_7*9.pdf) says that Clustered Singletons and Schedulers are intended to be deployed via farming (page 131 of 150). Why should I do this? Just to guarantee that I have the same versions across all nodes of the partition, or is there any other reason? I am deploying my Singletons and Schedulers through the default deploy dir and everything is working fine (almost).
I have implemented a Scheduler through Clustering. I have tested it on a partition with two nodes. Only one started as master, as expected. Than I removed the network cable from one of them. The slave started as master too. The problem arrised when I reconnected the network cable. I ended with two master nodes in a partition with connectivity. Is this the expected behavior? Shouldn't one of the masters stop and become a slave? Has any one experienced this? I am running jboss 3.2.3.
Thanks in advance,