By cmani on May 20, 2009
I was recently at a massive deployment of Sun Glassfish Enterprise server's (2.1) in a deployment, where the topology was the Glassfish DAS, Node agents, Sun web servers distributed across Solaris Containers across multiple servers - that is horizontal and vertical scaling.
After installing the DAS and Nodeagents, we created Glassfish server instances for all Nodegents, in an app server cluster. When we tried starting the app server instances, we started getting lots of "java.net.ConnectException: Connection refused exceptions" in the server logs and the servers would refuse to start. One of the clue's, of what the problem could be was in Shalini's blog. This led us to think, this was related to some kind of hostname issue. The problem, it seems that when you have a cluster of app server instances, there should be a hostname mapped to an IP address entry in /etc/hosts, for every app server instance participating in a cluster
So, in /etc/hosts, you should have something similar to the following:
xx.xx.xx.xx servername loghost
This problem, may probably not arise, if you there is a DNS server. In our case, we were operating from a Data center, with newly installed servers.