NFS server not responding/is alive again

[web at 3dresearch.com]

At 10:55 AM 5/3/2007, you wrote:
>Hi--
>
>Janos Dohanics wrote:
>>I have a FreeBSD 4.11 machine which mounts a volume from a Netapp ONTap.
>>The FreeBSD machine also acts as a Samba PDC. The Samba volumes are in
>>the NFS-mounted volume. There are about a dozen Win2K workstations on
>>the network served by the Samba server.
>>Lately I have noticed that /var/log/messages is full with entries like:
>>... /kernel: nfs server filer01:/vol/vol0/psa: not responding
>>... /kernel: nfs server filer01:/vol/vol0/psa: is alive again
>>It seems that the server sometimes is unresponsive for less than a
>>second, many other times it's unresponsive for a number of seconds (as
>>many as 8 seconds).
>
>In order to proceed, it would help to run a tcpdump between the NFS server 
>and this FreeBSD machine, and take a look at the packets just before one 
>of these errors is logged, and try to correlate with anything else in your 
>logs (ie, a particular Samba client did something or saw an error as a result).
>
>Use something like:
>
>   tcpdump -w packet.dmp -s 0 host filer01 and host localhost
>
>... and read via "tcpdump -r packet.dmp".

Chuck,

I did tcpdump -w packet.dmp -s 0 host filer01 and host localhost for a 
couple of minutes:

# tcpdump -w packet.dmp -s 0 host filer01 and host localhost
tcpdump: listening on dc0
^C
639 packets received by filter
0 packets dropped by kernel

... tcpdump -r packet.dmp gives zero output.

What does this tell you?

>Note that Samba is going to be happier serving from local disks; it would 
>be better for the clients to mount against filer01 directly than to 
>"forward" a remote filesystem via this NFS->CIFS/SMB bridge...

I'm sure you are right and I'd change it given the opportunity...

Janos