Are there problems with NFS version 4 clients mounting shares from NFS v3 servers?
I have a netapp array that is sharing files via NFS to solaris 10 servers. The netapp has both nfs v3 and v4 enabled. The solaris boxes were able to access these shares fine until the latest patch cluster was installed. Two days after the cluster was installed all of the sun boxes started having issues connecting to the nfs share and I see nfs errors dealing with stale sequence id numbers.
As a test we rolled the patch cluster back on one server but it still had the problem. We then forced the solaris systems to use NFS v3 and all of our problems are resolved; however, we really want to use NFS v4 on the solaris systems.
The network does not appear to be the issue. The network is very simple, hasn't changed, and even with little traffic we still have the problem on the v4 nfs client.
The Solaris implementation of NFS has always been very strict.
Other OS's? Not so much. Put an inode number larger than 32 bits while using non-large-file-aware NFSv2, like Irix would do? A Solaris host couldn't access the file. Nevermind some of the sketchy NFS sleight-of-hand Linux has been known to pull (silently truncating group membership?!?!?!).
And NFSv4 is pretty complex, and in my experience too easy to break.
I'd open tickets with Oracle and NetApp, and see if you can get something more out of them than pointing fingers at each other.