Return to BSD News archive
Xref: sserve comp.unix.bsd:16128 comp.protocols.nfs:11119 comp.os.386bsd.misc:5339
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msunews!uwm.edu!news.alpha.net!news.mathworks.com!newshost.marcam.com!charnel.ecst.csuchico.edu!olivea!nntp-hub.barrnet.net!nntp-sc.barrnet.net!netapp.com!netapp.com!not-for-mail
From: hitz@netapp.com (Dave Hitz)
Newsgroups: comp.unix.bsd,comp.protocols.nfs,comp.os.386bsd.misc
Subject: Re: NFS V3 Alpha release
Date: 11 Feb 1995 11:49:07 -0800
Organization: Network Appliance Corporation
Lines: 33
Message-ID: <3hj4bj$g50@supernova.netapp.com>
References: <3gm938$inb@ccshst05.cs.uoguelph.ca> <3gpq0s$5v2@news.cloud9.net> <3gu1o8$mf3@hpindda.cup.hp.com> <3hbcrp$dpt@engnews2.Eng.Sun.COM>
NNTP-Posting-Host: 192.9.200.16
In article <3hbcrp$dpt@engnews2.Eng.Sun.COM>,
David Robinson <robinson@jetsun.eng.sun.com> wrote:
>In article <3gu1o8$mf3@hpindda.cup.hp.com>, Rick Jones <raj@cup.hp.com> wrote:
>>How much of NFS V3 has to be in place before it can be called "NFS V3"
>>as opposed to "something that has many NFS V3 features?"
>
>My opinion is that the server must support all the ops in the V3
>specification that are appliciable to the native filesystem
>(exceptions like no symlinks allowed).
>
>The client side is more flexible because it may never issue some
>ops like mknod or symlink, but for the ops that it does generate
>they should be "correct".
I agree, but would like to add a more functional requirement:
o An NFS V3 client should be able to mount and access data
from an NFS V3 server. It is perfectly reasonable for a client
to choose not to use certain V3 features, such as READDIRPLUS
or COMMIT.
o An NFS V3 server should be able to satisfy requests from any
NFS V3 client conforming to the NFS V3 specification. Thus
it is *NOT* reasonable for a server to refuse to honor particular
NFS v3 requests. On the other hand, it would be acceptable for
a server to put all write data to disk immediately, and implement
the COMMIT operation by simply returning TRUE.
That might not be a *fast* NFS V3 implementation, but it would
certainly be legal.
Dave Hitz hitz@netapp.com
Network Appliance (415) 428-5106