Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 15 Jul 1997 08:15:22 -0400
From:      Dan Janowski <danj@3skel.com>
To:        freebsd-hackers@FreeBSD.ORG
Subject:   Re: multiple run-levels (was: Re: /etc/init.d/)
Message-ID:  <33CB69DA.9B1052@3skel.com>
References:  <Pine.BSF.3.91.970715103536.27554A-100000@ns.uk.peer.net>

next in thread | previous in thread | raw e-mail | index | archive | help
Hi All,

A multiple runlevel mechanism provides some benefits,
but most of them are secondary. The runlevel part is
mostly smoke.

The benefits are that each class or group of programs/processes
has its own shell script for initialization. This is a LOT like
the way we have stuff in /usr/local/etc/rc.d.

The other stuff, concerning what services should be started
if you are at init 2 instead of init 3 is more complicated than
what a "runlevel" can address.

A long time ago, when I was younger, I wrote an entire runlevel
and init.d/rc?.d system for Linux based on IRIX. There were only
two things that were of any use:

 1. At the boot loader prompt you could set a variable (LOCALE)
	that was available to the rc scripts that would allow
	the employment of location based control of what services
	should/shouldn't, network interfaces...

 2. chkconfig, which is an IRIX particular rc mechanism that allows
	one to type 'chkconfig timed off' at a prompt, and the
	next time you go to a multiuser state, timed would not be
	started. This meant that no actual file editing was necessary.

To REALLY be useful, a mechanism has to be able to change
resolv.conf, host.conf, hosts, ifconfigs, fstabs (for local and NFS),
libsocks.conf, routes, etc. All kinds of crap really. Init.d ain't gonna
do it. Another mechanism can be designed and implemented, but it
isn't really going to look like anything that we've seen before.

Dan


-- 
danj@3skel.com
Dan Janowski
Triskelion Systems, Inc.
Bronx, NY



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?33CB69DA.9B1052>