HACMP 5.1 Boot IP address issue 1

[spamly]

I've got a boot address on a standby node that isn't going away when the primary node's resource group fails to the standby node.

Here's how my cluster is setup:

I've got one cascading resource group in a HACMP 5.1 cluster. There are two nodes in this cluster which each have three interfaces. Both nodes are identical. Each node is configured with two standby intefaces and a boot inteface. The boot IP address on each node is public. When I startup HACMP services on the primary node, the boot IP address is taken down and the service (resource group) IP address takes its place. This is what I want.

Alright, the problem is that when the resource group fails over to the standby node, the standby node's boot IP address doesn't go away. Instead, the service IP address is bound to a standby interface. This is an issue because the boot IP address has the default route bound to it and the resource group needs to have any outgoing IP packets use its IP address, not the boot IP address.

Understand? What I need to know is how to configure the boot interface on the standby node to act the same as the primary node. I'd like the resource group to use the boot inteface and see that boot IP address go away. The standby interfaces should only be used in the event that the boot inteface fails.

I'm not using IP aliasing because the application doesn't support it.

Any ideas? Should I not use any public boot IP addresses? Maybe even stop using boot IP addresses all together and just have three standby IP interfaces configured on each node?

 

[DSMARWAY]

hi ,

in a cascading resource group the the standby card gets overwritten as you have experienced. thats how cascading works . whereas in a rotating resource group the boot address gets overwritten .

You can change your cluster to a rotating resource group
the benefit here is that you can bring the crashed box into the cluster without affecting the other node, but no node has priority as it does in a cascading resource group.

HTH