Implication of having ClusterService IP on same interface as another Service Group

[johno]

Because of a limited public interface adapter.  I placed the ClusterService
IP resource on the same device that I use for another Service Group.

ClusterService IP used NIC en0
SGA IP uses NIC en0

If I down the interface on NodeA to force a failover of the SG, it takes
upwards of 3 minutes to fail the IP address for Service Group SGA.  ClusterService
IP takes about 1 minute to 2 minutes.

If I disable the ClusterService SG then the Service Group SGA IP will fail
in about 10 to 30 seconds. 

Also what is the faileover time I should expect for a SG when the only rescources
involved are the NIC and IP resources. 

I am using Veritas Cluster 5.0 MP1
OS AIX 5.3 TL06
Interfaces are Virtual through the VIO server.

[RiaanBadenhorst]

Hi,

Why dont you just delete the Cluster Service group? I dont find it of much use.

Riaan

[SJP]

Hi,

You could delete the ClusterService Group but I think you'll find VERITAS would recommend against it for a number of reasons.

For example, the ClusterService group includes resources used to manage the Web-Based local Cluster Management Console. If you dont use the console then fair enough but its not the only thing that uses the ClusterService group if I remember rightly.

Might be worth checking it out with support, just to make sure.