Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!mintaka!spdcc!dyer
From: dyer@spdcc.COM (Steve Dyer)
Newsgroups: comp.unix.aux
Subject: NFS problems (was Re: A/UX concerns)
Message-ID: <6604@spdcc.SPDCC.COM>
Date: 25 Feb 91 19:27:23 GMT
References: <12183@goofy.Apple.COM> <1991Feb21.031529.4498@ni.umd.edu> <12191@goofy.Apple.COM>
Organization: S.P. Dyer Computer Consulting, Cambridge MA
Lines: 21

>BUT, I don't want people who are considering using or buying A/UX to get
>the idea that EVERYONE has trouble with NFS on A/UX: it just ain't so.

That's because not everyone operates in the same environment.  Just
because you haven't seen the problem doesn't mean it doesn't exist.
It's a problem with the Sun NFS sample implementation which most
vendors use more or less verbatim.  The problem is, under certain
conditions of server and network load, one or more of a series of
nonidempotent NFS requests which were scrambled (i.e. unserialized) by
biod may be retransmitted, and this can cause problems of 'holes'
appearing in files.  The loader seemed to be a particularly good way to
produce this, both with A/UX and with early versions of NFS for IBM AIX
PS/2.

We saw this at Athena early in our use of A/UX and reported it (along
with our sample fixes) many months ago.

-- 
Steve Dyer
dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
dyer@arktouros.mit.edu, dyer@hstbme.mit.edu