I have a JBOSS cluster setup with two servers load-balanced by an Apache web server with mod_jk/ajp13. Everything works fine with clustering and failover except in one situation.
When I need to bring a server down for maintenance the the server keeps getting requests for a few seconds while it's going down so these requests are lost to 503 and 404 errors until the load-balancer understands that it should not send any more requests to that node.
Is there any way to prevent this? Anything that I can do so that the load-balancer knows when a node is going down before the node will stop accepting requests?
Thanks in advance.