Help with creating an NFS proxy

[Emmanuel_Alao]

Hi Boys and Girls

I have got a 7 node cluster which has been configured with about 17 service groups, within each of these service groups there is an NFS type.
I was wondering in creating an NFS proxy so as to reduce the overhead of monitoring several NFS servcies do I create a seperate F/O service group
or should i just put it in the Cluster service group

[Gene_Henriksen]

You may get into trouble here with Proxy. Proxy is a persistent resource, therefore it has no offline routine and, by definition of a persistent resource, it should be at the bottom of a dependency tree.

Also, if you have a proxy, it could not start up the NFS it was looking for, only reflect its state.

Proxy works best with resources such as NIC and MultiNIC.

You did not mention the version of VCS or the OS.

I suppose you could put in a parallel group and have NFS in it (no Phantom resource required since NFS has an online routine). You could then have a Proxy in the SGs that need NFS and have it at the bottom or even unlinked but critical in the the SG.

If you put it in the ClusterService group it should only come online on one system as that is what CSG would do.

Test it first.

[Hywel_Mallett]

I've only used Proxy resources to proxy NICs.
You'd normally use a Proxy resource to replace multiple copies of the same resource.
I'm not entirely sure of the nature of NFS, but do you have exactly the same resource (with the same attribute values) in multiple service groups?
If so, then maybe Proxy is for you. If not (for example if each NFS resource is for a different NFS export), then I wouldn't see the point.

As Gene says, if the Proxy is what you want, then add a new service group with the NFS resource in it, and replace your NFS resources in your existing service groups with Proxy resources pointing to the new service group resources.

If you were doing this with NICs you'd need to add a Phantom resource to the new SG, as NICs are persistent resources, but for NFS you don't need Phantom.

[Emmanuel_Alao]

Sorry about that I am runnning VCS 4.1MP1 on node running solaris 8

[Emmanuel_Alao]

I have tried the parallel grp and also the failover grp what i find is that I can not get the service grp to come on line on any of the nodes but the type and nfs resource it self come on line on all the nodes

[Gene_Henriksen]

Then add a Phantom resource. Technically NFS is a persistent resource, but a special case because it does have an online routine.