SUMMARY: NFS timeout/delay problem

From: Dang, Vinh Khanh <Vinh.Dang_at_dsto.defence.gov.au>
Date: Wed, 06 Sep 2000 17:33:18 +1100

Hi,

Thanks to those who replied: Hoai Tran, Ian Mortimer, & Ralf W.
Grosse-Kunstleve.

It turns out that this wasn't a problem with the OS/NFS setup but rather
with the setup of the switches, that the fileserver and the client nodes
were connected to.

For those who are interested:
The ethernet ports on the fileserver/clients were configured to be 100Mbps
Full Duplex (set at the SRM console). The switches have an
'auto-negotiation' mode which is suppose to automatically configure the
ports on the switches correctly. However this 'auto-negotiation' mode set
the ports to be 100Mbps Half Duplex. Changing the switch port modes to be
Full Duplex fixed the problem.

Thanks again & regards,
Vinh Dang.


---- Original Message ----
Hi,

I have NFS setup on our fileserver and some client nodes. NFS appears to be
functioning correctly: the clients can mount and access the shared
directories on the fileserver, however occasionally, when a client node
tries to access the shared directories/files on the fileserver, the
operation (such as ls, df, cd, ...) freezes and the following message is
printed:

        NFS3 server fileserver_node not responding still trying

where fileserver_node is the name of the fileserver. Then after a while
(sometimes a couple of minutes), the following message is printed and the
operation is executed:

        NFS3 server fileserver_node ok

In the "Network Administration" book for Tru64, it says the first message
means "File operations in a hard-mounted file system have suspended because
communication between the client and the server has stopped" and the second
message means "File operations have resumed".

I tried increasing the number of NFS server daemons (UDP and TCP) to see if
this would make any difference, but it didn't. I then tried changing the
mount protocol from UDP to TCP, however this didn't make any difference
either.

Does anyone have any suggestions as to what is wrong?

Additional Info:
Fileserver and client nodes are running Tru64 4.0F.
NFS on fileserver is 'Configured for locking'.
NFS UDP server daemons = 32, TCP server daemons = 96.
Clients mount fileserver directories using an entry in /etc/fstab:
        fileserver_node:/mount/dir /mount/dir nfs rw,tcp,hard 0 0
And the mount command shows the entry as:
        fileserver_node:/mount/dir on /mount/dir type nfs
(v3,rw,tcp,hard,intr).

I would appreciate any suggestions or advice. Thanks in advance.
Regards,
Vinh Dang.
Received on Wed Sep 06 2000 - 06:40:22 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:41 NZDT