NFS problem

[ricky_8]

Hi ALL,

We have a NFS server setup on 11.x and clients on 11.x .It is new setup and iam facing some unusual problem.
I have installed latest PHNE_28102 patch on both server and cleints.Problem is i can mount NFS filesystem but i cannot "ls" on certain directories in it.It hangs and finally gives "NFS server not responding". At this exact moment i checked network with ping/telnet/ftp/rcp and all these services work to and fro from client and server.
So not quite sure whether itis patch issue or something else?

Any help will be highly appreciated.Thanxs,

[S.K. Chan]

If basic network connectivity is fine then this could be related to NFS performance. Check this thread for the NFS performance troubleshooting tips that I had attached, the doc is from technical knowledge db and it's specific to 10.x. Apart for some variation in the commands the basic concept stays the same.
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0xae40ec08252fd611abd50090277a778c,00.html
Beyond that one more thing you can do that may help. On the NFS server (the file /etc/rc.config.d/nfsconf) there is a var called "NUM_NFSD" and by default it's set to 4. Increase that number to say 8 (ie double it) to begin with. That will increase the nfs daemon on the server. You need to restart your nfs services for it to take effect. Then check ..
# ps -ef|grep nfsd
to make sure it now has 8 processes. On our server we increase that number to 12 and it does help (ie it reduced the "nfs server not responding" error). Do ..
# man nfsd
for details.

[ricky_8]

Hi,

The nfsd and biod are already set to 16.and the performance doc is not of much help in my case.
Is there something else i should look into.

[V. V. Ravi Kumar_1]

hi,
You are saying "Certain Directories" may be those directories are agian mount points on the server. check it out.

see the link
http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x6093afe90f1cd71190050090279cd0f9,00.html

Regards

[ricky_8]

Hello,

Unfortunately that is not the case. While searching logs i found following error on this server.
Apr 30 17:08:04 mentortc rpcbind: svc_tli_create: t_getname failed; T_IDLE

Does this ring bell to anybody?

[Ramkumar Devanathan]

Hi, are these nfs mounted disks present on a disk array or some other setup like that.

I've seen that to cause problems of permissions sometimes.

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x9b2d9607df6ed711abdc0090277a778c,00.html

see the link above. however i am sorry i am not able to help you. some more information/logs would help.

- ramd.

[Steven E. Protter · Accepted solution]

The most frequent cause of such problems are networking problems, having to do with duplex and speed.

If there is a mismatch, even one you can't easily see, it could cause this problem.

Lets say the NIC card is lan1 for this example.

lanadmin -x 1

It should and may say 100 BaseT Full Duplex.

If it says auto negotiate, you might want to change that to manual, especially if Cisco built your network switch. For some reason Cisco switches can confuse HP NIC cards.

Have the switch hard coded to 100 BaseT full duplex.

Boot your HP box.

lanadmin -x 1

It might suddenly show Half Duplex now.

lanadmin -X 100FD 1

To make the change permanent, add entries to the /etc/rc.config.d/hpbtlanconf file.

Here is a gsc card example.

/etc/rc.config.d/hpgsc100conf

HP_GSC100_INTERFACE_NAME[1]=lan1
HP_GSC100_STATION_ADDRESS[1]=
HP_GSC100_SPEED[1]=100FD

SEP

[Russell Campbell]

Are you seeing this problem when you cd to /net//.... and then do a ls -l?

[ricky_8]

Hi All,

Thanxs to steven's message which lead to check the mtu size for my gigabit card.The card was set to 9000 , when i turned it to 1500 mtu.

Evrything started working fine, but iam not sure why i was not able to make it work on 9000.

Anyfurther inputs onthis are welcome.

[KO kwang tae]

do you apply the following patch?
in case of hpux 11.11

PHCO_24777
PHNE_28103