NFS server not responding

[Mercadie]

Hi,

I have a L2000 HPUX 11.00 nfs client mounting a filesystem from a rp5405 HPUX 11.11 nfs server.

When creating, copying files (of any size) the following error is sometime logged:
Jun 20 17:47:30 asezat vmunix: NFS server wilsonpriv2 not responding still trying
Jun 20 17:47:30 asezat vmunix: NFS remove failed for server wilsonpriv2: RPC: Timed out
Jun 20 17:47:30 asezat vmunix: NFS server wilsonpriv2 ok

asezat is the name of the client and wilsonpriv2 the name of the second interface of the server

Has someone ever met this issue ?

[Steven E. Protter]

I have gotten this before.

It happened during intermittant network problems with wiring and the switch.

It happened when the target filesystem got full.

It happened when I crashed the nfs.core nfs.server daemon on the server.

SEP

[Massimo Bianchi]

Hi,
with NFS on HPUX is a very frequent problem.

If there are load issue or high load, nfs server ans client loose contact

Usually changing some mount parameter on the clinet side can help, but not resolve.


Try with these mount parameters:

mount -o rsize=32768,wsize=32768,hard server:/export /yourdir


where server:/export /yourdir are your true values.

HTH,
Massimo

[Elena Leontieva]

I think it ia probably caused by some network glitches. If you have your NFS server down (reboot) you also would see these messages but for a longer period of time. It also depends on how you configured NFS mounts.

[Steven E. Protter]

I have also seen these errors on servers that don't have all the necessary NFS patches. Thats the first place to look.

SEP

[Paul Sperry]

You either have network problems or you rebooted wilsonpriv2 and asezat has an nfs mount to wilsonpriv2 in which case this error is perfectly normal

[Caesar_3]

Hello!

Check you nfs server configuration.
Does your server serve many clients that have
lower verions than 11i like the asezat.

Also you may check the kernel configuration
of the client.

On http://docs.hp.com you can find pdf for
tunning the preformance of nfs.

Caesar

[A. Clay Stephenson]

You aren't going to like my answer but this is normal NFS behavior. NFS clients have to expect that servers are going to die at times.

This is not to say that you don't make your boxes and network as robust as possible. Apply the latest NFS/ONC/network patches, by all means. Also make sure that the network itself is as solid as possible. If this transfer is taking place over a WAN then problems are even more likely.

[Mercadie]

Thanks to All,

It seems the problem came from the configuration of the lan interface of the NFS server: It was 100 Full Duplex but with auto-negociation ON.

It seems that "sometimes" it was going to 100 half duplex. I don't know why because the client and the server are directly connected (cross cable) and the client interface is set to 100 Full duplex with auto-negociation OFF !

Now the interface of the server is set to 100 Full Duplex with no autonegociation and it works !

No more errors seen even after executing a loop that copy several big files.