From owner-freebsd-questions  Mon Sep 29 01:39:08 1997
Return-Path: <owner-freebsd-questions>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id BAA27789
          for questions-outgoing; Mon, 29 Sep 1997 01:39:08 -0700 (PDT)
Received: from gatekeeper.barcode.co.il (gatekeeper.barcode.co.il [192.116.93.17])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id BAA27769
          for <freebsd-questions@FreeBSD.ORG>; Mon, 29 Sep 1997 01:39:00 -0700 (PDT)
Received: (from smap@localhost) by gatekeeper.barcode.co.il (8.8.5/8.6.12) id KAA24222; Mon, 29 Sep 1997 10:35:46 +0200 (IST)
X-Authentication-Warning: gatekeeper.barcode.co.il: smap set sender to <nadav@barcode.co.il> using -f
Received: from localhost.barcode.co.il(127.0.0.1) by gatekeeper.barcode.co.il via smap (V1.3)
	id sma024219; Mon Sep 29 10:35:22 1997
Message-ID: <342F684B.7F67@barcode.co.il>
Date: Mon, 29 Sep 1997 10:35:23 +0200
From: Nadav Eiron <nadav@barcode.co.il>
X-Mailer: Mozilla 3.0 (X11; I; SunOS 5.5.1 sun4u)
MIME-Version: 1.0
To: Greg Lehey <grog@lemis.com>
CC: Donn Miller <dmm125@bellatlantic.net>, freebsd-questions@FreeBSD.ORG
Subject: Re: "df" estimates are off
References: <Pine.NEB.3.96.970929011000.28104B-100000@myname.my.domain> <19970929153303.59702@lemis.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-questions@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

Greg Lehey wrote:
> 
> On Mon, Sep 29, 1997 at 01:13:06AM +0000, Donn Miller wrote:
> > Hello,
> >
> > Recently, I've noticed that the df command has been showing negative disk
> > space, and was wondering how far negative it could go until the /usr
> > partition was full:
> >
> > Filesystem  1K-blocks     Used    Avail Capacity  Mounted on
> > /dev/wd0a       48431    39551     5006    89%    /
> > /dev/wd0s1     423072   392984    30088    93%    /dos
> > /dev/wd0s4e    548911   526867   -21868   104%    /usr
> > procfs              4        4        0   100%    /proc
> >
> > But then I noticed that, for /usr, it shows -21868 avail, but when doing
> > the math, 22044 is actually available.  For /, shows 5006, but act. free
> > space is 8880.
> 
> ufs allows users access to only 90% of the total space on disk.  The
> rest is reserved for root.  To make it more confusing, the 90% mark
> can be changed with tunefs(8), so the designers chose to refer to
> whatever this mark is as 100%.  Under normal circumstances, this means
> that root can pump the file system up to 111% of its nominal
> capacity.  This is *bad*, especially on /usr, since no normal user can
> write to the disk any more.

And then there is the performance issue. This margin is designed to
allow the filesystem to wisely place files on disk (to avoid DOS-style
fragentation). By filling it up you also loose performance.

> 
> > The disk usage for /dos is right on target.
> 
> This suggests that /dos is not ufs.
> 
> > Is it because df is allowing for a 'safety margin' of free space, or do I
> > need to reboot or rebuild some database or something?
> 
> You need to remove some data, or you'll run into trouble.
> 
> Greg
Nadav