[Mailman-Users] Managing lists as a non-root/Mailman user

David Southwell david at vizion2000.net
Thu Apr 19 15:41:41 CEST 2007


On Thursday 19 April 2007 06:05:38 Steve Burling wrote:
> --On April 18, 2007 9:55:48 PM -0500 Brad Knowles <brad at shub-internet.org>
>
> wrote:
> > In the nearly twenty years I've been doing this sort of stuff, I don't
> > think I've ever ran into a single NFS environment that I would call
> > "well-managed".  Some were not as badly managed as others, but they all
> > had major problems.
>
> To which I reply:
>
> That's to bad.  In the nearly thirty years that I've been doing this sort
> of stuff, I've come across a few.  I work in (and help manage) one now.

well if I remember correctly NFS was developed by Sun but not first introduced 
fully until  very early spring of 1989 (RFC 1094). However the degree of time 
over which any one of us has been associated, to one degree or another, with 
its use, is IMHO, far less significant than the facts that have emerged in 
practice. The evidence does seem to point pretty conclusively towards the 
notion that what might be interpreted as a "well-managed" environment depends 
very heavily upon flow-control conditions. This is a factor that is not under 
the total control of administrators. 

NFS dynamics over flow-controlled wide area networks by Chang, Morris & Kung 
of Harvard ( INFOCOM '97) is recomended reading. The evidence clearly shows 
that NFS over TCP works badly with small packets but is pretty much OK over 
ATM WANS if the flow control keeps the cell loss rate under 1%.

NFS is like anything else - it reflects not how long we have been doing it for 
but how well we perform in the currently prevailing conditions!!

David


More information about the Mailman-Users mailing list