Subject: Re: Proposed rc.d changes....
To: None <email@example.com>
From: Darren Reed <firstname.lastname@example.org>
Date: 05/01/2000 07:52:42
In some email I received from Luke Mewburn, sie wrote:
> Darren Reed writes:
> > Something eles I'd like to be able to do, which would take support
> > from init(8), is to be able to do "shutdown now" from my xterm and
> > have it go to single user mode with the console redirected to my
> > xterm (or at least not kill the xterm & shell).
> So, first of all you argue against the shutdown keyword as it's (to
> paraphrase you) `useless creeping featurism', and then you want
> something which is non trivial, reeks of `useless creeping featurism',
> and and is really bound to be useful to a smaller subset of users
> (i.e, `you').
Well, it's not something I'm going to parade up and down about asking
for. I think HP-UX will behave this way, and when you're sitting at
your workstation with your servers in the basement, being able to do
"shutdown" from your X-term and keep that shell as interactive in
single user rather than walk down to an uncomfortable cell which is
cold and isoalted, does have its advantages. Anyway, I thought I
would just mention that as a nice feature to have since the discussion
was about shutdown behaviour and will quite willingly back down from
supporting it if it means doing without the extra "shutdown".
> E.g, in the majority of cases running '/etc/rc.d/network stop' at
> shutdown time loses for people. However, some people have expressed
> a desire to be able to stop and start the network manually. Thus,
> the network script could support `stop' for manual control, but have
> `shutdown' as a no-op.
Ah, I see.
If we had something which did "S*" and "K*" scripts, there would be
an "Snetwork" so people could delete "Knetwork" and adding "shutdown"
is the answer to the "remove the K-script" problem (in a round about
fashion). If "network stop" is a problem because it happens too early,
then maybe that needs to be addressed separately (for example, it should
not stop just because you go to single user mode). As much as this
makes me puke, unless we went to a different implementation of rc, there
is no other way around this problem. If we had a tool to manage all of
the rc system, then I'd say here should be a checkbox which said "run at
shutdown [ ]" which would cause "shutdown" to invoke "stop" for that
script if selected. Maybe this is a sign that we suck when it comes to
rc :-/ (don't take that personally)