Problem with NFS over TCP

[Heinrich Rebehn <rebehn@xxxxxxxxxxxxxxxxx>]

Hi list,

while i was making some NFS performance measurements, i stumbled into this:

Server: GragonFly BSD
Client: Linux 2.6.7

root@antsrv1 [/home/data] # grep nfs /etc/rc.conf
netfs_types="nfs:NFS smbfs:SMB portalfs:PORTAL nwfs:NWFS" # Net filesystems.
nfs_client_enable="NO"          # This host is an NFS client (or NO).
nfs_client_flags="-n 4"         # Flags to nfsiod (if enabled).
nfs_access_cache="2"            # Client cache timeout in seconds
nfs_server_enable="YES"         # This host is an NFS server (or NO).
nfs_server_flags="-u -t -n 16"  # Flags to nfsd (if enabled).
nfs_reserved_port_only="NO"     # Provide NFS only on secure port (or NO).
nfs_bufpackets=""               # bufspace (in packets) for client

root@antsrv1 [/home/data] # cat /etc/exports
/home   -alldirs        -maproot=0      compsrv1 compsrv2 compsrv3

*TCP:*

root@compsrv1 [~] # mount|grep antsrv1
antsrv1:/home/data on /mnt type nfs 
(rw,v3,rsize=32768,wsize=32768,hard,tcp,lock,addr=antsrv1)

root@compsrv1 [~] # time dd if=/dev/zero of=/mnt/`hostname`.dat bs=1024k 
count=1000
dd: writing `/mnt/compsrv1.dat': Input/output error
217+0 records in
216+0 records out
dd: closing output file `/mnt/compsrv1.dat': Stale NFS file handle

real    0m21.327s
user    0m0.006s
sys     0m2.035s


*UDP:*

root@compsrv1 [~] # mount|grep antsrv1
antsrv1:/home/data on /mnt type nfs 
(rw,v3,rsize=16384,wsize=16384,hard,udp,lock,addr=antsrv1)

root@compsrv1 [~] # time dd if=/dev/zero of=/mnt/`hostname`.dat bs=1024k 
count=1000
1000+0 records in
1000+0 records out

real    1m37.962s
user    0m0.015s
sys     0m10.096s


The error does not happen when the server runs FreeBSD-5.3, so i don't 
think that the Linux NFS client is at fault.

There are no messages in /var/log/messages

root@antsrv1 [/home/data] # netstat -m
339/1427/26624 mbufs in use (current/peak/max):
        339 mbufs allocated to data
338/1091/6656 mbuf clusters in use (current/peak/max)
2538 Kbytes allocated to network (12% of mb_map in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines


Any idea what to tune or how to debug?
--

Heinrich Rebehn

University of Bremen
Physics / Electrical and Electronics Engineering
- Department of Telecommunications -

Phone : +49/421/218-4664
Fax   :            -3341