Subject: Re: space the final frontier [was sysinst report]
To: Robert.V.Baron <firstname.lastname@example.org>
From: Simon J. Gerraty <email@example.com>
Date: 12/06/1997 13:54:47
>As I suggested before, a 32Meg root and 128Meg usr and the rest usr1
>sounds good to me.
Hmmm, most machines I install these days (Solaris,NetBSD whatever)
use a single fs for /,/usr and /opt on Solaris I use a 600M root!
Never had a minutes trouble with any of them, OS upgrades etc are
a breeze etc. And minimizes wasted disk space. Big fs's like that
are made (by me anyway) with 1% slack rather than newfs's default of
10%. If /tmp is going to be a memort fs, then swap is 200-600M depending
on the system.
For NetBSD I use 400M if /var is going to be in /, otherwise 200M.
For machines that are exposed and need lots of logs I use a
separate /var of 1-2G.
SunOS used to install an 8M!!! root on SunOS 4.0 and it was virtually
impossible to fit more than 1 kernel in it. 64M sounds a safer minimum
than 32M - I hate having to waste time shuffling files around because
filesystems are full.
Most of my machines these days get /,/share and typically a single mount
point for each additional disk.
There is a PR raised against 1.3ALPHA because the install dies if you
don't allocate a separate /usr btw.