NFS problems ! TimeOut.

[Chapaya]

Hello ,

I get this messages when i'm using NFS :

NFS write failed for server XXX: RPC: Timed out

Eran .

[Massimo Bianchi]

Hi,
usually this is due to heavy network traffic.


To avoid this, you can change mount parameters.


Try with this:
hard,rsize=32768,wsize=32768

Massimo

[Michael Tully]

You could increase the number of background nfsd processes that run. You can do this is in the /etc/rc.config.d/nfsconf file. One other thing to look at is the network itself. Make sure that you are running 'full duplex' and not half duplex.

# lanadmin -x 0 (assuming lan 0 is your card)

to change it
# lanadmin -X 100FD 0

[Chapaya]

Hello ,

Both interfaces are 100FD .

Eran.

[RolandH]

Hi Eran,

first do a nfsstat on your client site
# nfsstat -c

have you a lot of timeouts in your statistics?

Then see the rsize, wsize from your nfsmounts with
# nfsstat -m
With HP-UX 10.20 ist was 8192 for rsize and wsize. Since HP-UX 11.00 it is per default 32768.
So if you have 11.00 or above and a lot of timeouts in you statistics, then you should set the rsize and wsize parameter down to 8192.

Try this first.


HTH

Roland

[Massimo Bianchi]

Hi,
you may also think of installing deviceid,
it saved me a couple of times.


http://www.software.hp.com/cgi-bin/swdepot_parser.cgi/cgi/displayProductInfo.pl?productNumber=DEVICEID

Massimo

[Chapaya]

Hi Roland ,

What is the purpose of rsize & wsize ?

Eran .

[RolandH]

Hi Eran,

this are the packet size in bytes.

r=read
w=write

I have had the problem couple of weeks ago.
The nfs network traffic from one city to the other failed always with the same error message as yours. Or it taken a lot of time to transmit data from one to another system over the nfs link with very small files. These was all 11.00 and above. With the 10.20 systems there was no problem. After reducing the rsize and wsize to 8192 level it works.


Roland

[Chapaya]

Hello ,

I have this problem just with big files ~ 1G .
The copying proccess takes too much time and after a while the copying process is aborted .

Eran.

[Michael Tully]

Are the two systems on the subnet?

[Chapaya]

Yes ,they are in the same subnet .

[RolandH]

Ok Eran,

I think we need more info from your system.
( C=Client, S=Server)
What OS do you have? (C + S)
What NFS version do you use?
# rpcinfo -u nfs
# rpcinfo -u nfs

Give us your nfs statistics (C + S)
# nfsstat -cnrs

Give us your statistics for every mounted file system (only from C)
# nfsstat -m

How looks your /etc/rc.config.d/nfsconf ?

Have you installed the latest ONC/NFS General Release/Performance Patch?

10.20 PHNE_22288
11.00 PHNE_28567
11.11 PHNE_28568
( all require reboot)

How looks your /etc/fstab ( only the nfs mounts) or your manually mount command?

Have you tested with decreased rsize wsize?

Roland

Roland

[RolandH]

Hi Eran,

I found also this Document: NETUXKBRC00006283
in the Technical Knowledge Base

Thera are a lot of hints to your problem.

Roland