In message <7245.1060249861@munnari.OZ.AU>, you write: 

->     Date:        Wed, 06 Aug 2003 13:26:35 -0400
->     From:        Rafal Boni <rafal@pobox.com>
->     Message-ID:  <200308061726.h76HQZji015435@fearless-vampire-killer.waters
-> ide.net>
-> 
->   | 	Anybody out there have good experience running NFS over 802.11? 
-> 
-> What system version?

Relatively -current... I can't recall if I updated that machine last I
updated the rest of my NetBSD boxen, but it should at the very least be
running 1.6T (and if I upgraded it too, it's running a at-most-week-old
-current).

-> For the wi at least, if you're running a system without the improvements
-> David Young made a few months ago, then I'd expect to see problems with
-> NFS [...]

Hmm, it's definitely newer than a few months.

-> There's no reason I can think of why NFS (in particular, as opposed to the
-> traffic patterns it generates) would work better or worse over one kind of
-> (approximately) 10Mbps medium than any other (it isn't as if the drivers
-> or hardware care one way or the other).   

True, though I suspect that unlike other 10Mbps media, 802.11 loses quite a
bit more often (packets, that is :-), and I would guess more succeptible to
data-dependent drops (ignoring the higher variability in both raw bandwidth
and delays in real world situations).

It's either that, of the cards have so little buffering that back-to-back
packets lose even under good circumstances... I say this because in the
case of "hung" NFS mounts, I see a lot of "frags dropped after timeout"
in the IP stats (I hadn't looked there the first time until sending off
the email made me think about fragmentation-related foo).

--rafal

----
Rafal Boni                                                     rafal@pobox.com
  We are all worms.  But I do believe I am a glowworm.  -- Winston Churchill