By monitoring CPU and RAM utilization, it appears that one member of your cluster is receiving a much lower percentage of the request load than the other server...