[NCLUG] NFS from HPUX to Linux

[Mike Loseke]

Thus spake "JONES,LUKE (HP-Greeley,ex1)":
> 
> I've got a couple of Linux boxes from which I would like to mount a
> filesystem exported off an HP-UX server. It's not working.
> 
> Since I have two Linux boxes, and since I don't know the first thing about
> HPUX, I thought I would start by checking my understanding of NFS. I was
> able to get things working there but only by disabling the firewalls on both
> boxes.
> 
> (Side note: the NFS howto anticipates this, noting that you can specify
> particular ports for /(lock|quota|stat|mount)d/ but if anyone has done this
> with the delightful Red Hat value-add networking scripts (e.g. a secret
> environment variable to set in /etc/sysconfig/network, or something like
> that) I'd appreciate it if you could pass that information my way.)
> 
> So, anyway, I can do both an export and a mount with Linux boxes and it's
> quasi-straightforward to get that working. But the HP-UX box RPC times out
> on me. Teh PHXU box shows a bunch of plausible services when i do rpcinfo -p
> (nfs, mountd, etc.) either on it or remotely, so I'm puzzeld what to make of
> tihs.
> 
> Any idesa, PHXU mavens?

 I've had no problems mounting HP-UX B.11.00 and B.10.20 machines from my
linux clients. Currently all linux machines here are defaulting to NFSv3/tcp
and negotiating down and working totally fine across all NFS-mounted
systems.

 Here's some info that may or may not help:

 example HP-UX 11 /etc/fstab entry:

 /dev/vg01/lvol1 /export/foo vxfs rw,suid,nolargefiles,delaylog,datainlog 0 2

 rpcinfo -p output:

 linux-macheen$ rpcinfo -p hpux-macheen
   program vers proto   port
    100000    4   tcp    111  portmapper
    100000    3   tcp    111  portmapper
    100000    2   tcp    111  portmapper
    100000    4   udp    111  portmapper
    100000    3   udp    111  portmapper
    100000    2   udp    111  portmapper
    100024    1   tcp  49153  status
    100024    1   udp  49159  status
    100021    1   tcp  49154  nlockmgr
    100021    1   udp  49160  nlockmgr
    100021    3   tcp  49155  nlockmgr
    100021    3   udp  49161  nlockmgr
    100021    4   tcp  49156  nlockmgr
    100021    4   udp  49162  nlockmgr
    100020    1   udp   4045  llockmgr
    100020    1   tcp   4045  llockmgr
    100021    2   tcp  49157  nlockmgr
    100068    2   udp  49165
    100068    3   udp  49165
    100068    4   udp  49165
    100068    5   udp  49165
    100083    1   tcp  49158
    100005    1   udp  49171  mountd
    100005    3   udp  49171  mountd
    100005    1   tcp  49163  mountd
    100005    3   tcp  49163  mountd
    100003    2   udp   2049  nfs
    100003    3   udp   2049  nfs


 Make sure that your HP-UX machine is running the protocols that your
client is asking for. HP-UX has always done the NFS thing just
different enough to cause issues like this. Default r/w sizes was their
biggest departure from the 'norm'. These are more normal now.

 On the HP-UX machines, the NFS system reads it's config out of
/etc/rc.config.d/nfsconf. It's mostly self-documented with references to
man pages. The only real changes made to mine are to pass additional
variable definitions to the automounter (which is client, not server).

 tcpdump or snoop (solaris) is your friend when troubleshooting NFS
connections between machines. Look for version discrepancies and tcp/udp
mismatches. You may have to tune your client mounts to match what the HP-UX
server wants to serve.

 But, if at all possible, construct a trebuchet large enough to hit the
HP plant where you are and launch that pile of junk back at em and use
Linux for your NFS services (or a nice big fat RAID system). :-)

-- 
   Mike Loseke    | Eagles may soar, but weasels
 mike at verinet.com | aren't sucked into jet engines.