NFSv4Support over Solaris 10

[Robur]

In VCS 5.1 SP1 Administration Guide (Solaris), NFS Service Group configuration, paragrah "Sample configuration for a single NFSv4 environment" mentions 'NFSv4Support' parameter for this NFS version support and lock recovery, but this parameter doesn't appear in types.cf. Any idea about this?

On the other hand, is it really necesary a DNS resource for supporting NFSv4 as indicated in these pages?

Thanks in advance.

[mikebounds]

This is very odd. I have seen quite a few times where Symantec leave information only relevent to old versions in guides, but I have looked in VCS bundled agents guide for 4.1, 5.0, 5.0MP3, 5.1, 5.1SP1 & 6.0 and the NFSv4Support attribute does not exist, so I don't understand where this has come from.  The only reference I can find to NFS v4 in the bundled agents guide is in the Mount agent where it says:

For IPv6 functionality for NFS, you must use NFS version 4 in order to make

the mount reachable. AIX defaults to NFSv3, which does not work across

IPv6. Note that NFSv4 requires several configuration steps in the operating

system and NFS-related resources in VCS to enable it on the client and the

exporting server.

Also, as you say, I don't know why they show a DNS resourse as you only normally need DNS resources if you have nodes (in an replicated data cluster or global cluster) where the IP fails over to one in a different subnet.  The sample configuration shown is not in a global cluster and the IPs are not localised in the IP resource (as you would do in an RDC) so DNS resource should not be required.  However, the vcs admin guide does say:

Create a DNS resource if you want NFS lock recovery. The DNS resource

depends on the IP resource. Refer to the sample configuration on how to

configure the DNS resource.

But it does not explain why.  I have successfully recovered NFS locks before without DNS resources, but this was with NFSv3, so I don't if NFSv4 is different for some reason.

Mike

[Robur]

Thanks Mike for your response.

In addition...

This VCS 5.1 cluster is a Preproduction one that has 2 NFS Service Groups but in Production cluster we can have more than 10 NFS Service Groups. Documentation recommends to include two (2) NFSRestart resources (lower and upper) in every NFS SG, so in case of switching/failing over one of these NFS SG (and according to this documentation), nfs daemons (nfsd, mountd, statd, lockd) would be rebooted/refreshed in that node and that doesn't seem to us a good idea because ALL NFS file systems look like "frozen" and applications can't evolve normally.

Is there any supported configuration that permits NFSv4 lock recovery without having nfs daemons rebooted?

[Venkata_Reddy_C · Accepted solution]

Thanks for bringing this up. NFSv4Support attribute is applicable only on Linux OS.

That was a documentation bug for Solaris. We will get that corrected in the example.

I will get back to you on the more specific NFSv4 sample configuration for your use case next week.

Thanks,

Venkat