NFS stops responding

Benjamin Scott dragonhawk at gmail.com
Fri Apr 2 11:18:05 EDT 2010


On Fri, Apr 2, 2010 at 11:03 AM, Michael ODonnell
<michael.odonnell at comcast.net> wrote:
> Heh.  I'd love to, and I just acquired a brand new switch to use as
> an experimental replacement for the one currently deployed.  I'll be
> ecstatic if that fixes thing, though I'm not optimistic.

  Me neither, but I suspect that's a good thing to try.  Easy to try,
and sometimes weird stuff happens.

> ... the problem is not that we stop seeing return
> traffic from the server, it's more that the client code stops making sane
> decisions in response when it arrives.

  Where/how are you running Wireshark?  I would suggest running it on
the NFS client, on the NFS server, and on a third computer monitoring
the link.  If the switch(es) you have do(es) not support port
monitoring (AKA mirroring AKA other things), find an old 10/100
repeater and stick that between client and switch, then server and
switch.

> Maybe the packets aren't getting
> all the way back down the stack to be processed by the client code?

  I was wondering the same thing.  Unfortunately, my knowledge of NFS
is limited.  I haven't used NFS in years, I haven't used it seriously
in over a decade, and I never really got involved in diagnostics when
I was using it seriously.  (I do remember that NFS had the unique
property of being able to bring down an entire network of machines
just by shutting down one server, if the cross-mounts were gnarly
enough.  Granted, that was with the implementations on ancient SunOS
boxes and an early Linux NFS, both of which were notoriously fragile.)

-- Ben



More information about the gnhlug-discuss mailing list