Having exiting node reboot after cluster failover

[John Theune]

I have a application that is not cluster aware and when a failover of
the group occurs one of the application programs does not release a
network resource. What I would like to do is have the node hosting
the group reboot itself after the failover, so that it releases
everything. I thought about putting a generic app in the group and
having it call a batch file to reboot the other node when the group
first starts up. This issue with this is I need to make this generic
for several hundred different clusters and I don't know the node
names. Is there a way to have a batch file run on the exiting node
during the failover or can I set the group to reboot the other node
during group startup?

 

[Guest]

John,

If node 1 fails and node 2 assumes ownership of the groups from node 1, you probably have to assume that there is no reason to think that node 2 will be able to issue any commands to node 1, because the failover wouldn't occur unless node 2 stopped communicating with node 1 in the first place. So you can't rely on the cluster service on node 2 to fix your problem.

Your best bet would be to isolate the reboot issue to node 1. The simplest way would be to script an AT task that checks to see if the cluster service on node 1 is running. If the response comes back as no, send a reboot command on the local node.

Sean Bogan
Microsoft Consultant
Collective Technologies
www.collectivetech.com