Christos Zoulas wrote:

>On Jan 4,  9:04pm, riz@redcrowgroup.com (Jeff Rizzo) wrote:
>-- Subject: Re: NFS problems
>
>| I actually just ran into a problem like this between 2 NetBSD 2.0 (i386) 
>| systems today... NFS over UDP, where client accesses just started 
>| hanging until I rebooted the client.  (although, "shutdown -r now" did 
>| successfully reboot for me).
>| 
>| My processes weren't zombies; ps showed them in disk wait;  all other 
>| network traffic seemed OK, but NFS just died.  (after some days of being 
>| fine, and it's fine again after client reboot)
>| 
>
>Try decreasing the read and write sizes on your mounts.
>
>christos
>  
>
Hm.  I can certainly give this a try (they're at the default now), 
though that will make it hard to determine for certain why the problem 
doesn't reappear (assuming it doesn't).  I'm not really convinced it's 
related, because one of the things I noticed was that while the problem 
was occurring, an ls in one of the nfs-mounted directories did not 
generate _any_ network traffic.  (a 'df' generated some fsstat calls, 
but still hung)

I think at the moment, I'll wait for it to happen again, and gather as 
much info as possible.  Any particular suggestions? Incidentally, the 
ethernet on both boxes is:

fxp0 at pci0 dev 4 function 0: i82559 Ethernet, rev 8
fxp0: interrupting at ioapic1 pin 4 (irq 11)
fxp0: Ethernet address 00:e0:81:03:0d:00
inphy0 at fxp0 phy 1: i82555 10/100 media interface, rev. 4

+j



-- 
Jeff Rizzo                                         riz@redcrowgroup.com
Red Crow Group LLC                                 http://www.redcrowgroup.com/
+1 415 550 0310