We are having an issue where we have 4 nodes setup in a cluster, and under a small load it works fine and processes requests. When the load is increased to a certain point, we are seeing that the java processes for each node have their cpu utilization drop to 0 and no further requests are processed.
When this happens, the database server is not slow and is at a very low cpu utilization. The same can be said for the webserver.
If i run with only one node in the cluster, i don't have any issues. But with two nodes, i can recreate the problem.