Guys,
Let me first give you the background of the configuration:
Two node VCS4.1 cluster on Solaris 10. Each node connected to two subnets:
subnet A: 169.187.46.0/24 (primary one with default GW)
subnet B: 169.184.147.24/29
Traffic to some subnets I need to pull thru subnet B. In subnet B there is no HSRP or any other virtual GW that would failover between physical routers but only physical routers that I can use as GW. So on one node I route traffic thru .25 and on the other the traffic goes thru .26.
I have a NIC resource (only one interface is set up on subnet B on each node, no ipmp due to HW limitations in very small DMZ network). NIC has a NetworkHosts variable within which we can define hosts to ping. Ideally would be to ask VCS to monitor NIC by pinging .25 on one node and .26 on the other node and make sure that if the GW we use on the particular node becomes unavailable VCS will switch the group. VCS does not allow this and we can only define same IPs that can be used on both nodes.
Any ideas how I can solve it with VCS? Obviously I would like VCS to pick up GW failure that is used on the node and swap the group to the second node on which we use the other GW. I am thinking of setting up and application resource that would start a simple script. The script would be pinging the GW and would exit if the GW turns out to be unpingable two or three times in raw. If there is only one ping failure the counter would reset to zero back again.
Let the brainstorm begin...
Thanks in advance.
Wojtek
