David,
This is an interesting problem. You left out considerable detail from the NFS Perspective, but then not many of us are NFS / Automount experts. What I recommend that you do is use the typical "bottoms up" approach to determining what your network link(s) is(are) capable of first.
Since you're on J6000, verify the speed and duplex values for the lan card(s) match those expected with the switch. Use lanscan to get the PPA for each link, and then lanadmin -x
to get the speed/duplex for each link.
If you have any mismatch, or further network config/debugs for network links, I highly recommend you look at http://techsolutions.hp.com/
I also HIGHLY recommend using TTCP or Netperf to measure the throughput capabilities of the network. FTP is not a valid test tool in spite of what you may think, for a variety of reasons. Use TCP or UDP, for this test, depending on the type of NFS connection in use, and specify the TCP or UDP packet size 8K or 32K depending on the size of the NFS requests (look on the client nfsstat -m output for the mount point in question)
TTCP at http://www.faqs.org/faqs/hp/hpux-faq/section-213.html, netperf at http://www.netperf.org
Once you've gotten past this, you're into NFS Performance-land, for which we have two VALUABLE references.
1. NFS Server Not Responding ... vmunix: NFS write failed for server ... - WTEC a comprehensive document
Now available in Kmine/ITRC customer viewable document NETUXKBRC00006283
2. NFS Performance Assessement ITRC/Kmine Document KBAN00000261 Excellent writeup for NFS Performance issues!
3. NFS Performance and Tuning Presentation - Dave Olker's Customer Presentation on 11.0 11.i http://devresource.hp.com/devresource/Docs/Presentations/NFSperf.pdf
Hope all this helps,
-> Brian Hackley
Ask me about telecommuting!