From: "J. Bruce Fields" Subject: Re: [PATCH 2/2] nfsd: default to kernel default for minorversion 1 Date: Fri, 5 Feb 2010 11:10:02 -0500 Message-ID: <20100205161001.GB24926@fieldses.org> References: <1264631166-21898-1-git-send-email-bfields@citi.umich.edu> <1264631166-21898-2-git-send-email-bfields@citi.umich.edu> <20100201195821.GB19418@fieldses.org> <4B6B480C.1050307@RedHat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-nfs@vger.kernel.org To: Steve Dickson Return-path: Received: from fieldses.org ([174.143.236.118]:51086 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754062Ab0BEQJp (ORCPT ); Fri, 5 Feb 2010 11:09:45 -0500 In-Reply-To: <4B6B480C.1050307-AfCzQyP5zfLQT0dZR+AlfA@public.gmane.org> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Thu, Feb 04, 2010 at 05:19:56PM -0500, Steve Dickson wrote: > On 02/01/2010 02:58 PM, J. Bruce Fields wrote: > > On Wed, Jan 27, 2010 at 05:26:06PM -0500, J. Bruce Fields wrote: > >> The current kernel code should not be enabled by default, because it > >> does not yet attempt to be a conform completely to the rfc; for example, > >> some required pieces of protocol are missing. > >> > >> Therefore the kernel defaults to leaving minorversion1 off. When the > >> code matures sufficiently, that default will change. > >> > >> That kernel default becomes meaningless if nfs-utils always explicitly > >> turns 4.1 on or off. So, nfs-utils should by default do nothing. > >> > >> Provide a --enable-experimental-v41-support option to turn it on > >> explicitly. The option is intentionally spelled out (and has no short > >> equivalent), to help ensure that users know what they're getting into. > Command options like this are so hard to get rid of.... We just can't introduce > an option one release and then have it go away a few releases down the road. > That's sure fire way to breaking existing configurations which something that is, > has been and will continue to be unacceptable... OK, fair enough. It shouldn't be a problem to keep it indefinitely, though. > >> Once 4.1 defaults to on, that option will become unnecessary (and can > >> probably just be dropped from nfs-utils), and only the -N 4.1 option > >> will be necessary. > > > > We need to figure out how we're going to handle this. The current > > situation (ignoring the kernel's default) isn't acceptable. > Why can't each distro simple turn it off with there init scripts? On the server-side the kernel doesn't have a separate config option for V4.1; the only way for the kernel to indicate whether it has a mature version of V4.1 is by its choice of default. > > We need to figure out something for v4 as well. If every distro has run > > nfsd without -N4, depending on the lack of fsid=0 to keep nfsv4 off by > > default, then we're effectively turning v4 on by default with the > > pseudoroot changes. But there are good reasons why the v4 server code > > is still marked experimental. > I agree that with the latest kernels, v4 will be the default version. But > there are ways to override this on both the server and client. So why let > the users decided what they want? I don't understand. --b.