I am back again with NFS troubles.
I have configured two machines : one as NFS server (both 2.2.5-ac1) using
knfsd 1.2, and one NFS client.
The tests :
o I am running iozone (200Mb file with 8192 block size) from
the client on a file on the server.
One occurence works pretty well.
Two concurrent occurences of iozone on two different files
fail randomly for one occurence with :
> Writing the 200 Megabyte file, 'zz3'...Error writing block 6788
> iozone: Erreur d'entrée/sortie.
Usually, the other iozone completes successfully.
In the client log file, I have :
> nfs: server willy not responding, timed out
o Now, I am running iozone from my Linux box, and iozone from
one or more Solaris or FreeBSD (for example) clients.
Most of the time, the iozone on the Linux box fails again
with the same messages. All other iozones complete successfully.
So, my conclusion is that the Linux box is rather sensible to NFS
time-out.
The mount parameters on the Linux client are :
willy:/u0 on /a/willy.u0 type nfs (rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1,vers=2,addr=134.10.44.51)
The mount parameters on the Solaris clients are :
/willy/u0 on willy:/u0 vers=2/remote on Wed Mar 31 12:48:54 1999
I remember that Trond proposed once to increase NFS verbosity to tracks
things, but I do not find his message.
What could I do to help ?
A+,
-- Thierry Danis danis@sagem.fr- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/