NFS SERVER NOT RESPONDING

[abhijit chaphekar]

I have three machines in a cluster viz zek,quark and nog. Zek is a nfs server and other 2 are clients.

Direct auto mounting method is used for exporting file systems.

Till date the nfs was functioning fine.But now we have problem with one filesystem i.e. "/var/mai" which when tried to access from machine "quark" then the machine hangs giving "NFS SERVER NOT RESPONDING" ERROR.
While the same filesystem gets mounted on other client "nog" without any sort of problem.

If we use Standard mount on "quark" for /var/mail then it works.

If we remove the entry for /var/mail from automounter map file for machine "quark" then it works fine for all other automounted filesystems.

Please can any one help me in this matter.
ThanX in Advance,

[Stefan Farrelly]

Take a look at the following link for a good summary;

http://www.dutchworks.nl/htbin/hpsysadmin?h=3&dn=16902&q=nfs%20server%20not%20responding&fh

[Chris Renshall]

First of are the clients and the server at the same patch level.

I find hpux automounter a little querky anyway and mostly don't use itexcept for /net .

What I do is use SAM and add all my mounts as straight NFS no automounts and I find far less trouble this way.

My system consists of a J2240 actign as a Master NFS server with a smaller 715 as a slave, these service about 70 C class stations with no problems.

Good luck

[Rick Garland]

Can run the command '/usr/bin/nfsstat' to see if there might be something causing the block. Will provide stats on the performance of the mounts as well.

[Shannon Petry]

The problem could be a couple of things. I disagree with the mention of patch levels being the same. They have nothing to do with each other!
If you can mount it manually, then the fstab entry, or automounter should work too.
The most common error I see is "access denied" when HP hosts can not resolv a host name, so this probably is not it.
First, make sure that the export file has the correct access. Also make sure that the server can resolv the clients name properly, and visa-versa. If any restrictions are set on mounting, then this is critical.
Next, check the NFS server to see that it has enough NFS daemons running to handle all of it's tasks. If there are many mounts, and few daemons then it could take very long to connect.
If you use SAM on the client to mount, then aliases are corrected if possible, and you should be warned of unknown hosts and invalid shared directories. A lot of the time, a simple typing error can cause the biggest headaches!

Best regards,
Shannon

[abhijit chaphekar]

Chris,
As you pointed out the patches are not same on all these machines.Iam not sure about which patch might be causing the problem here is the patch list attached with this mail.
further inputs will be helpful.

[Anthony Goonetilleke]

Have you changed your kernel parameters recently? There used to be a problem where Autmount would fail if maxfiles > 2048 but
PHNE_21376 fixes that problem.