GoldenGate Server collector processes continue to run on failed RAC node

I have an issue with Best Practice Knowledge Article 1313703.1. I have set this up in a RAC environment and when I failover from one node to the other, the GoldenGate server collector processes ("./server") continue to run on the failed node. The article above does not mention nor does it handle cleanly and transparently shutting down the collector processes on the failed node. Those processes remain connected to the remote GG server and will not stop unless the data pump on that server is first terminated. I have modified the action script to find the server collection processes and