From owner-freebsd-hackers  Sun Jan  4 21:24:24 1998
Return-Path: <owner-freebsd-hackers@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id VAA03567
          for hackers-outgoing; Sun, 4 Jan 1998 21:24:24 -0800 (PST)
          (envelope-from owner-freebsd-hackers@FreeBSD.ORG)
Received: from fang.cs.sunyit.edu (root@fang.cs.sunyit.edu [192.52.220.66])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id VAA03462
          for <hackers@FreeBSD.ORG>; Sun, 4 Jan 1998 21:23:36 -0800 (PST)
          (envelope-from perlsta@sunyit.edu)
Received: from win95.local.sunyit.edu (ppp-55.ts-6.nyc.idt.net [169.132.98.55]) by fang.cs.sunyit.edu (8.8.5/8.7.3) with ESMTP id BAA29931 for <hackers@FreeBSD.ORG>; Mon, 5 Jan 1998 01:24:20 GMT
Message-Id: <199801050124.BAA29931@fang.cs.sunyit.edu>
From: "Alfred Perlstein" <perlsta@sunyit.edu>
To: <hackers@FreeBSD.ORG>
Subject: Re: Musing on boot
Date: Mon, 5 Jan 1998 00:19:47 -0500
X-MSMail-Priority: Normal
X-Priority: 3
X-Mailer: Microsoft Internet Mail 4.70.1161
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk

this is a great idea, also the ability to add something along the lines of:

if (mount on not_so_important_fs fails) sendmail > root "why it failed"

also a syslog or something mechanism for saving the fsck output would be
nice...
maybe it exists but i can't ever find it... kinda like something to log the
startup script output....

-Alfred

----------
> From: Warner Losh <imp@village.org>
> To: hackers@FreeBSD.ORG
> Subject: Musing on boot
> Date: Sunday, January 04, 1998 11:40 PM
> 
> 
> I've hacked my rc files so that I always fsck on boot.  This worked
> out fairly well until recently.  Recently, I added about 6G of space
> across several partitions and disks.  Now the system takes forever to
> boot.  I had thought about unhacking the fsck, but then I realized
> that it would take forever to boot when I've crashed.
> 
> So, being the safety conscious impatient engineer that I am, I thought
> about implementing the following.  I thought I'd bounce it off hackers
> first to see what people think.
> 
> The idea is to have a list of file systems that *MUST* be present for
> the system to come up.  These files systems are fsck'd and mounted
> synchronously.  All the rest of the file systems have a fsck kicked
> off in the background, and a mount done when that fsck happens to
> finish.
> 
> This is horrible for home directories, but great for the OpenBSD
> sources, FreeBSD source, NetBSD sources, /usr/obj, build trees,
> gcc/egcs expermental crap, etc that is scattered over much of the new
> disk space.  This would allow me to get back up quickly, while
> allowing stuff to "drift" into the system as it is available.
> 
> I thought I'd bounce it off hackers.  It seems like such a simple idea
> that something must be wrong with it.
> 
> Comments?
> 
> Warner
> 
> P.S.  Code to follow if there appears to be interest...