NFS performance issue

[Sean Roe]

Hi All,

I am rtf'ing like mad but I cant seem to find an answer to my problem.  I 
have an nfs server running Debian 3.1 and every once in a while, the load 
goes through the roof (like 20+). Its an duel hyperthreaded intel 
processor based machine that just hums along (it doesn't know the words) 
most of the time.  But every couple of minutes or so the load spikes and 
the nfs becomes unresponsive. Here is output from one of the clients:

Nov  3 15:52:39 ws01 kernel: nfs: server nfs01 OK
Nov  3 15:54:22 ws01 kernel: nfs: server nfs01 not responding, still 
trying
Nov  3 15:54:23 ws01 kernel: nfs: server nfs01 not responding, still 
trying
Nov  3 15:54:41 ws01 kernel: nfs: server nfs01 OK
Nov  3 15:54:41 ws01 kernel: nfs: server nfs01 OK

Here is the nfsstat from the server:

Server rpc stats:
calls      badcalls   badauth    badclnt    xdrcall
-1309384875   33375      0          33375      0
Server nfs v2:
null       getattr    setattr    root       lookup     readlink
0       0% 1334410941 90% 2074877  0% 0       0% 133796690  9% 516     0%
read       wrcache    write      create     remove     rename
1766674  0% 0       0% 5275039  0% 889651  0% 211550  0% 659132  0%
link       symlink    mkdir      rmdir      readdir    fsstat
0       0% 6       0% 1752    0% 354     0% 14094   0% 9       0%

Server nfs v3:
null       getattr    setattr    lookup     access     readlink
17      0% 1210394011 80% 1601356  0% 42514827  2% 60271952  4% 3632    0%
read       write      create     mkdir      symlink    mknod
89763759  5% 74699132  4% 574313  0% 15226   0% 636     0% 0       0%
remove     rmdir      rename     link       readdir    readdirplus
128993  0% 486     0% 48753   0% 1       0% 155881  0% 2399    0%
fsstat     fsinfo     pathconf   commit
585     0% 580     0% 0       0% 26304597  1%


I am not really sure how to read this but it looks reasonable to me.

Anybody got any helpful insight to this?

Thanks,
Sean