What are the best practices to shut-down a hung SOA-P node during it's processing existing requests from remote JMS server?

Issue

• There are 3 node SOA-P servers running cluster talking to a 2 node JMS server (e.g. using JBoss Messaging stack) cluster running on remote JBoss EAP servers. The ESB applications deployed on SOA-P are dependent upon the JMS server to be up and running. That means before shut-down of SOA-P server takes place someone has to shut down SOA-P servers first, then the JMS server instances. Is there a way to break that dependency? Perhaps it could be achieved by shutting down certain services on SOA-P server instances so it stops listening on the remote queues?
• The current problem is if the JMS server goes down, SOA-P server instances can't be shut down because it will be hung waiting to connect to the JMS queues, and it's process needs to be killed.
• What are the best practices to shut-down a hung SOA-P node during it's processing existing requests from remote JMS server?