NFS client error : NFS write failed for server

NFS client error : NFS write failed for server

Post by hagai yaf » Wed, 28 Jan 2004 01:08:13



Hello.

   I am using a Solaris 8 machine and working with an NFS remote file
system, when trying to copy big files to the NFS file system I am
getting the next errors :

NFS write failed for server shlight: error 5 (RPC: Timed out)
NFS write failed for server shlight: error 5 (RPC: Timed out)
NFS write failed for server shlight: error 5 (RPC: Timed out)
cp: /export/home/prod/import/a.tar: write: Connection timed out

I have a slow connection to the NFS server, I think that this is
causing my problems, is there a way to tell the NFS client on my
Solaris machine to use a bigger time out before aborting the operation
with en error ?

TX.
    hagai.

 
 
 

NFS client error : NFS write failed for server

Post by Michael Tos » Wed, 28 Jan 2004 02:36:19



> Hello.

>    I am using a Solaris 8 machine and working with an NFS remote file
> system, when trying to copy big files to the NFS file system I am
> getting the next errors :

> NFS write failed for server shlight: error 5 (RPC: Timed out)
> NFS write failed for server shlight: error 5 (RPC: Timed out)
> NFS write failed for server shlight: error 5 (RPC: Timed out)
> cp: /export/home/prod/import/a.tar: write: Connection timed out

> I have a slow connection to the NFS server, I think that this is
> causing my problems, is there a way to tell the NFS client on my
> Solaris machine to use a bigger time out before aborting the operation
> with en error ?

> TX.
>     hagai.

See
man mount_nfs

Promising is e.g.

mount -o proto=udp,retrans=10,timeo=600,rsize=8192,wsize=8192,retry=20000
...

Maybe one of these is already enough, try it out!

--
Michael Tosch
IT Specialist
HP Managed Services Germany
Phone +49 2407 575 313

 
 
 

NFS client error : NFS write failed for server

Post by Casper H.S. Di » Wed, 28 Jan 2004 04:33:58



>   I am using a Solaris 8 machine and working with an NFS remote file
>system, when trying to copy big files to the NFS file system I am
>getting the next errors :
>NFS write failed for server shlight: error 5 (RPC: Timed out)
>NFS write failed for server shlight: error 5 (RPC: Timed out)
>NFS write failed for server shlight: error 5 (RPC: Timed out)
>cp: /export/home/prod/import/a.tar: write: Connection timed out

Don't mount with "soft".

Quote:>I have a slow connection to the NFS server, I think that this is
>causing my problems, is there a way to tell the NFS client on my
>Solaris machine to use a bigger time out before aborting the operation
>with en error ?

In my experience with soft mounts the timeouts are never long enough.

So use hard mounts and your troubles are gone.

"hard,intr" (the default) does work well for mounts like these: you can
still bail out (intr) but you won't get corruption (hard).

Surely you don't want the mount to be both slow *and* unreliable?

Casper
--
Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

 
 
 

1. Poor NFS write performance from Linux clients to NFS Ver 3 servers.

I'm having a real problem with very slow NFS write performance from Linux
clients running Redhat 4.2, 5.0, and 5.1 to servers running NFS version
3.0.  The servers indicate that the Linux clients are mounted using NFS v2.
The servers I have are running Solaris 2.5.1 and 2.6 on both Intel and SPARC
hardware. Others have reported the same problem with Alpha servers running
Digital UNIX and NFS v3. My servers are all up to date on patches. The
network (Fast Ethernet) is not the problem as I have this problem on two
isolated networks at different locations. NFS read performance is normal.

I have increased the rsize and wsize to 8192, which doubled the performance,
and have played with every option I thought would help.  Now the writes are
only a factor of 10 times slower than FTP'ing the file and 6 times slower
than other clients running Solaris 2.5.1 and NFS v3 using TCP.

I'm getting desperate as everyone is screaming about the slow performance.
I'm hoping that someone has solved this problem. Surely I don't have to put
extra disks of home directories on these Linux boxes to be able to use NFS.
I just installed Redhat 5.1 hoping that would fix my problem, but its no
different. Why is it taking so long to add NFS version 3.0 protocols to
Linux?  It has been available from other vendors for well over a year. The
performance of version 3.0 using TCP is significantly faster than version
2.

Thanks in advance for any help. I will post a summary if I find a solution.

Denny Morse

2. Problem with command 'join'

3. NFS server on HP-UX 10.20 & NFS client on RH-5.2 : write permission ???

4. samba problems.

5. NFS backup auto mount on client when NFS server fail...!

6. Installing i386 Redhat4.1 from HD

7. Need info on PC/NFS and Unix NFS servers, server/client ratio

8. dual boot? win 95/5.0.4 - how do you do it?

9. How do I debug NFS write error on host ABC: Stale NFS file handle?

10. Awful NFS client perf. w/AIX NFS server

11. Tracing NFS Clients on the NFS Server ?

12. SCO NFS client with HP/Apollo NFS server?

13. problem with LINUX (NFS server) and Solaris 2.7 (intel ) NFS client.