Custmoised agents(resources) and service groups are not working corrrectly

[Sridhar_sri]

Hi friends,
                 I have created a customised agent for primary server.When one of the process goes down in primary server , the agetn and its dependent resource is going down in primary server. where as the cluster service group which has been created during Veritas cluster configuration responsible for Notifier manager, Veritas cluster management console, is not getting offline completely instead it is getting back to ONLINE in primary server itself.
                    Due to this thought the rest of the resources are coming to ONLINE in secondary server , cluster service group are not online in secondary srvfer, the reources and service group are getting faulted and returning back to primary server itself.

                    Can any one help me to solve this issue

Thanks,

Sridhar

[M__Braun]

Hi Sridar,

Could you please provide some details:

- How are the Service Group configured?

- Which resources are in the Service Group?

- Why is the cluster service group affected by your Custom Agent?

- Can you provide main.cf excerpts from the relevant SGs?

Regards

Manuel

[Gaurav_S]

Hi Sridhar,

Please add more clarity on your query..

a. You are saying you have created an agent for primary server ? Is agent not on secondary server ?

b. As Manuel asked above, have you linked your data service group with ClusterService service group ? If yes, what relationship it is ?

Gaurav

[TomerG]

I'm not sure here, but I think Sridhar may be confused. The ClusterService group is independent of your application service group, and doesn't need to be affected. If your application service group fails over, why do you think the ClusterService group should? It controls cluster infrastructure items, like notification. It shouldn't need to failover unless there has been a system fault. 

[Eric_Hennessey1]

This almost sounds like the resource being managed by your custom agent has been placed in the ClusterService group.  If that's the case, you should move it out of there and into its own separate service group.  As Tomer mentions above, the ClusterService group is a special service group intended for cluster infrastructure management.