Strange NFS problem (not responding still trying)

[Saiful Muhd Ramli]

Hi all,

I'm seeing a really strange NFS issue, some directories from our NFS servers consistently bring up the mesg "NFS server agmxsv18 not responding still trying" while other directories from the same server do not..
Even wierder this is happening on only one system, all other systems seem to connect fine..

I get the mesg after I cd to the directory and then do an ls.. Even wierder some child dirs exhibit the same problem while others do not...!

Client h/w is an HPUX 700i running 10.20 and autofs.

Any suggestions??

[Steven E. Protter]

Quite strange.

Perhaps check the /var/adm/syslog/syslog.log on both systems.

Also, check kernel parameters. Maybe nproc, nfile or maxuprc is causing issues.

SEP

[Ralf Buchhold]

Hello
is the ping of the nfs host ok ?
nslookup ok ?
look at resolv.conf for dns

By
Ralf

[Saiful Muhd Ramli]

Steven,

Here's a stupid question, how do I know if
the kernel parameters are causing problems?

Ralf, ping and nlookup work w/o fail. In
fact, what is strange is that the failure
seems to be in which dir on the server I try
to do "ls" on.. Certain ones work and others
fail, consistently.. All along, nslookup
and ping works..

[V. Nyga · Accepted solution]

Hi,

have you checked lan speed of your ws and of the switch it's connected to?

Volkmar

[Ian Hillier]

Check your /etc/exports file on your NFS server and make sure the client hostname (or IP) is listed as having access to that share. If you make any changes to the file, run exportfs -a after you save it. This will not affect any of your other NFS clients.

Regards,

John

[Fred Ruffet]

You should maybe try to increase your NFS processes :
NUM_NFSD=16
NUM_NFSIOD=16
this variables can be increased in /etc/rc.config.d/nfsconf.
You'll have to restart NFS server/client then.

Regards,

Fred

[Ashwani Kashyap]

If the client that is having this problem automounts the directories , then I think it is a normal behavior because unused directories will be unmounted by the automounter and vmunix will complain about the nfsserver npt responding . but if you cd into the directory you will still be able to see all of its contents because automount will remount it .

As far as the child directories behaving strangely , it could be a confiuration problem in they are mounted specifically if they are indirect mounts . Check your configurations .
It s been a long time since I worked with autmounter .

[Phillip Renner]

I don't know about 10.20 but we've had this issue with 11i systems and we had to remove the timeo value from the fstab and that fixed it. I have a nfs performance doc from hp that says the timeo overrides the Van Jacobson algorith and can degrade system performance and cause the NFS not responding pages. Sooooo, maybe increase the value or run without it if 10.20 allows..

[Bill Hassell]

NFS is a fairly fragile protocol so if your network is busy and uses a lot of hubs, you should consider scrapping all hubs and use switches. Run nfsstat and post the results. The stats will tell you where you're havig problems.