From owner-freebsd-isp  Mon Aug 26 13:53:44 1996
Return-Path: owner-isp
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id NAA15143
          for isp-outgoing; Mon, 26 Aug 1996 13:53:44 -0700 (PDT)
Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.109.160])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id NAA15121
          for <freebsd-isp@freebsd.org>; Mon, 26 Aug 1996 13:53:40 -0700 (PDT)
Received: (from jgreco@localhost) by brasil.moneng.mei.com (8.7.Beta.1/8.7.Beta.1) id PAA01720; Mon, 26 Aug 1996 15:52:05 -0500
From: Joe Greco <jgreco@brasil.moneng.mei.com>
Message-Id: <199608262052.PAA01720@brasil.moneng.mei.com>
Subject: Re: Anyone using ccd (FreeBSD disk striper) for news
To: nate@mt.sri.com (Nate Williams)
Date: Mon, 26 Aug 1996 15:52:04 -0500 (CDT)
Cc: jgreco@brasil.moneng.mei.com, nate@mt.sri.com,
        michaelv@mindbender.serv.net, freebsd-isp@freebsd.org
In-Reply-To: <199608262017.OAA20405@rocky.mt.sri.com> from "Nate Williams" at Aug 26, 96 02:17:54 pm
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-isp@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> > Phahff.  Screw POSIX compliancy if it's an optional brokenness and it makes
> > life better.  DG has mentioned that it's a real problem for him on wcarchive
> > in the past, too...  and there are some of us who understand the need for
> > standards compliance but also appreciate that there are times that the rules
> > can be safely bent.  :-)
> 
> UnionFS would be a *really* good solution in this case.  You'd allow
> someone to have 'read/write' priviledges to the FS (inn, maintainers,
> etc..), but then re-mount it somewhere else read-only, thus disabling
> ATIME writes and only allowing read-only FS's.

I don't know about UnionFS...  but in the name of thoroughness let me go
build a kernel..

Damn.  Instant crash.

The comments in LINT do mention that it is buggy.

> The best of both worlds.  Otherwise, I think if you added a flag to the
> FS to disable ATIME updates for specific filesystems you might get DG to
> add it. (hint hint! ;)

We discussed it in the past, both DG and I have come up with implementations
that we were not happy with for one reason or another (me, my solution was
probably naive because I have zippo familiarity with FFS internals, DG's
solution probably had technical correctness on it's side but for some reason
didn't work in all cases, or something like that).

... JG