Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Mac OS X Mail 6.6 \(1510\))
Subject: Re: [PATCH 3/4] NFSD: Add WRITE_PLUS support for hole punches
From: Chuck Lever <chuck.lever@oracle.com>
In-Reply-To: <20131104185309.GA14378@infradead.org>
Date: Mon, 4 Nov 2013 11:16:13 -0800
Cc: Anna Schumaker <bjschuma@netapp.com>, bfields@fieldses.org,
        linux-nfs@vger.kernel.org
Message-Id: <3AC4ECC6-0496-4593-8791-92BCAD2567CC@oracle.com>
References: <1382972247-1108-1-git-send-email-bjschuma@netapp.com> <1382972247-1108-4-git-send-email-bjschuma@netapp.com> <20131102135238.GB18961@infradead.org> <5277CE23.6010207@netapp.com> <20131104170345.GA31499@infradead.org> <5277D80F.8050106@netapp.com> <20131104185309.GA14378@infradead.org>
To: Christoph Hellwig <hch@infradead.org>
Sender: linux-nfs-owner@vger.kernel.org


On Nov 4, 2013, at 10:53 AM, Christoph Hellwig <hch@infradead.org> wrote:

> On Mon, Nov 04, 2013 at 12:23:27PM -0500, Anna Schumaker wrote:
>>> state" that will return zeroes when read, even if those zeroes didn't
>>> make it to disk.
>> 
>> And that's all done through metadata?  Doing the commit_metadata() call makes a bit more sense now.
> 
> Yes.  The only filesystem in tree that actually seems to write zeroes
> is gfs2, but even then it doesn't do so through the usual mechanisms.
> 
>>> Also what does the data arm buy us over good old WRITE?
>>> 
>> 
>> I've been working off of draft #21, the most recent commit is 2b3ab1740b1ea843faa59566fb4213a42d8c724a from Aug 19.  The eventual goal is to phase out WRITE and replace it with WRITE_PLUS, but that won't happen until at least 4.3.  I don't know why it's being done this way instead of just adding an FALLOCATE operation.
> 
> Where can I find draft 21?  The newest document in the git repo I was
> pointed to earlier is draft-ietf-nfsv4-minorversion2-13.txt, and the
> newest I could find on tools.ietf.org is draft 20.
> 
> I still don't understand why anyone would phase out WRITE in favour of
> something that doesn't actually add any value for the write case.

Protocol extensibility.  WRITE_PLUS adds a discriminated union of data types that can be extended easily to include initialization patterns, integrity metadata, and other things like holes.  It's entirely another question as to whether any particular extension is tasteful.

> I'll try to take it up directly with the working group, but my post to
> the list yesterday in your SEEK thread didn't seem to have made it
> trough.   I also tried to research how they came up with this idiotic
> design, but the mailing list archives tell very little.  Could it be
> that most of these decisions are actually made in a smokey backroom and
> not on the list?

The documents themselves count as public proposals.  That's why they are referred to as "Requests For Comments".

The mailing list is not the only place where they are discussed.  We also use conference calls, and there are face-to-face meetings too.  Not all of that content is distilled into a public record.  Very much like Linux kernel development.

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com