I have two clustered jboss nodes for this example I will call them A and B.
If i start node B first and then node A everything works fine and the servers start up in <45 seconds. If I switch the order I have problems. First problem is that it takes Node B FOREVER (like 5 minutes) to get past the [service] Removing bootstrap log handlers. When it finally continues on it gets to the setting the clustered beans to the HA JNDI and it takes 1-2 minutes ber bean. Meanwhile on node A I continously get ERROR [org.jboss.system.server.profileservice.repository.clustered.ClusteredDeploymentRepository] getModifiedDeployments() Cannot acquire local lock and then line above says Unable to acquire lock as it is held by Node B] Even with those error messages everything seems to be functioning just a lot slower
e.g. typical tran for us take 300 milliseconds when this happens they take around 2000 milliseconds
Any help is greatly appreciated.
To anybody that this was affecting I found the problem. The issue was somewhere in the farming deployment we were deploying the application in the farm directory for all the servers and for some reason one order in starting them up worked and another way didn't. We now just deploy the code in the deploy directory and everything works great!