Requests are being forwarded directly from BigIP to Appserver. We dont have any applications which have session persistance.
Secondly, the way bigIP checks if an appserver is up or not is by accessing a static URL which we have deployed as an application on each cluster member. Static URL produces a string either "Server UP" or "Server Down ". If the response from Static URL is not "Server Up", big marks the JVM down. In a cluster, there could be multiple JVM's , so i am trying to check just bigIP URL rather than going to each individual JVM and checking for its status.
BigIP URL :
abc.com/checkJVM.jspIf the response from URL is "Server UP" , i wouldnt like to stop cluster but if the response from URL is "Server Down, then i would like to bring down the cluster. I have wsadmin scripts which do stopping/starting clusters.
The reason for my approach is basically avoid any human errors. Let me give a simple example- We have applications across datacenters.Sometimes during deployments, we come across situations where have livetraffic being handled by one datacenter ( ex: Datacenter A ) and we are supposed to deploy application on other center (ex: Datacenter B) and by mistake we end up stopping traffic on datacenter A causing an outage.
Please let me know if this gives an idea at what i am trying to accomplish.