I am running three jboss4.0.3sp1 on windows 2003 as a cluster
After some times unexpected windows OS shutdown on one machine(let's say C), the jboss server on C automatically startup again. But when it starts [TreeCache] service, some unexcepted instances appear. For example,
[TreeCache] viewAccepted(): A:1234, B:2345, C:3456, C:5678, C:5679, C:6788, C:6789
C:5678, C:5678 C:6788, C:6789 should not be there!! And YES!! They are the survivors of the last two times server shutdown!!
But I use netstat -ano to check the ports. They are not used at least in the cmd console view. So I guess they are fake instances.
For tomcat clustering(tc5-cluster-service.xml), I use only <FD_SOCK> protocol to do failure detection. It's default.
How can this happen?? And how to avoid this when unexpected windows shutdown can not be avoided at the moment(It happens actually rarely).