freebsd stable - Dec 2003 - NFS serer weirdness

Hello

I am running a STABLE NFS server as follows

9:28am mdouhan @ [rabarber] ~ > uname -a
FreeBSD rabarber.internal.hasta.se 4.9-STABLE FreeBSD 4.9-STABLE #3: Sat Dec 
20 17:25:35 CET 2003     root@rabarber.internal.hasta.se:/usr/obj/usr/src/
sys/RABARBER  i386
9:28am mdouhan @ [rabarber] ~ >

From this server I mount  NFS filesystems from a STABLE machine as follows

9:29am mdouhan @ [mandarin] ~ > uname -a
FreeBSD mandarin.internal.hasta.se 4.8-STABLE FreeBSD 4.8-STABLE #0: Sat Apr  
5 17:07:20 GMT 2003     root@mandarin.internal.hasta.se:/usr/obj/usr/src/sys/
MANDARIN  i386
9:29am mdouhan @ [mandarin] ~ >

using the following line in /etc/fstab

9:29am mdouhan @ [mandarin] ~ > cat /etc/fstab | grep rabarber
rabarber:/net/rabarber  /net/rabarber   nfs     tcp,rw,nfsv3    0       0
9:29am mdouhan @ [mandarin] ~ >

I also mount this NFS filesystems on a CURRENT machines as follows

9:30am mdouhan @ [ananas] ~ > uname -a
FreeBSD ananas.internal.hasta.se 5.1-CURRENT FreeBSD 5.1-CURRENT #0: Wed Nov 
19 19:55:50 CET 2003     root@ananas.internal.hasta.se:/usr/obj/usr/current/
sys/ANANAS  i386
9:30am mdouhan @ [ananas] ~ >

using the following line in /etc/fstab

9:30am mdouhan @ [ananas] ~ > cat /etc/fstab | grep rabarber
rabarber:/net/rabarber  /net/rabarber   nfs     tcp,rw,nfsv3            0       
0
9:31am mdouhan @ [ananas] ~ >

all machines are using the fxp NIC at 100Mbps fdx operations, none of the 
machines show any input or outpput errors if I check with netstat -i, they 
are all connected to a single Switch and the switch does not report any 
errors on the ports and also shows that correct speed/duplex has been 
negotiated, I have also tried hardcoding the speed/duplex settings but the 
result is the same

The problem is as follows
9:34am mdouhan @ [mandarin] ~ > tail /var/log/messages
Dec 19 11:53:37 mandarin /kernel: nfs server rabarber:/net/rabarber: not 
responding
Dec 19 11:53:37 mandarin /kernel: nfs server rabarber:/net/rabarber: is alive 
again
Dec 20 21:54:35 mandarin /kernel: nfs server rabarber:/net/rabarber: not 
responding
Dec 20 21:55:49 mandarin /kernel: nfs server rabarber:/net/rabarber: is alive 
again
Dec 20 22:04:24 mandarin /kernel: nfs server rabarber:/net/rabarber: not 
responding
Dec 20 22:04:38 mandarin /kernel: nfs server rabarber:/net/rabarber: not 
responding
Dec 20 22:05:48 mandarin last message repeated 2 times
Dec 20 22:06:04 mandarin /kernel: nfs server rabarber:/net/rabarber: is alive 
again
Dec 20 22:06:04 mandarin last message repeated 3 times
9:34am mdouhan @ [mandarin] ~ >

All machines mounting NFS systems are seing these errors in the log, and it is 
very annoying, anyone have any idea why this is?

Regards

Matt


-- 
Matt Douhan
www.fruitsalad.org
matt@fruitsalad.org
ping elvis
elvis is alive
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: signature
Url :
http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20031221/fa7af824/attachment.bin

On Sun, 21 Dec 2003, Matt Douhan wrote:
> I am running a STABLE NFS server as follows
[...]

Is it different if you use tcp?

Try using this command to mount the filesystem:

mount_nfs -T -3 rabarber:/net/rabarber  /net/rabarber

Then try your operation and see if it works better.  Also see if you
experience random hangs with this.

You may be having corruption/loss issues with small packets. You'll have
to run tcpdump and try to see where the packet loss is occuring.

-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite@gumbysoft.com          |  www.FreeBSD.org

On Sun, Dec 21, 2003 at 09:35:31AM +0100, Matt Douhan
wrote:
> The problem is as follows
> 9:34am mdouhan @ [mandarin] ~ > tail /var/log/messages
> again
> Dec 20 22:04:24 mandarin /kernel: nfs server rabarber:/net/rabarber: not 
> responding
> Dec 20 22:04:38 mandarin /kernel: nfs server rabarber:/net/rabarber: not 
> responding
> Dec 20 22:05:48 mandarin last message repeated 2 times
> Dec 20 22:06:04 mandarin /kernel: nfs server rabarber:/net/rabarber: is
alive
> again
> Dec 20 22:06:04 mandarin last message repeated 3 times
> 9:34am mdouhan @ [mandarin] ~ >
> 
> All machines mounting NFS systems are seing these errors in the log, and it
is
> very annoying, anyone have any idea why this is?
> 
> Regards
	This is an historical problem with dynamic retransmit
estimator. There is previous disscussion about this in the archives.
If everything is working OK you can ignore the messages or
mount with -d. see man 8 mount_nfs.

	-steve