How to prevent mod_cluster as single point failure?-Collection of common programming errors

In my configuration i used Apche+mod_cluster as a front_end(load balancer) server and two JBOSS AS 7.1.0 as backend server.

In my configuration on one system i have install Apache+mod_cluster + JBOSS AS 7.1.0 and on other system JBOSS AS 7.1.0. Both JBOSS AS 7.1.0 are in domain mode means in clustering.

My problem is that if Apche+mod_cluster crash than client can not access any JBOSS AS .

I want to configure Apache+mod_cluster in Master_Slave format, means if one of Apache+mod_cluter down another Apache+mod_cluster become active and pass client(browser) request to back-end server in normal form without any interruption.

Is there any way to make two Apache+mod_cluster to communicate with each other , means to check health status of each other and if any one get down other Apche+mod_cluster will do it’s task ……

1. One of the many options is a load-balancing using round-robin DNS, which can also serve as a “poor-man’s fault-tolerance”

   Put 2 hosts with apache+mod_cluster in front of appservers. Let’s call them apache-1 and apache-2. Add IP’s of both of them to the same DNS record.

   You don’t need any master/slave configuration, once the next request achieves apache-2 instead of apache-1, it will proxy it to any appserver behind:

   1. If the chosen appserver has the user session, it is served,
   2. Otherwise the session is found across the cluster, migrated to currently selected appserver and served