Subject: Re: CVS commit: basesrc
To: Luke Mewburn <>
From: Tim Preston <>
List: current-users
Date: 05/16/2001 17:37:15
Luke Mewburn wibbled one day
> On Wed, May 16, 2001 at 10:02:23AM -0400, Thor Lancelot Simon wrote:

> > > There's nothing stopping you from hitting ^C to interrupt the wait
> > > whilst rc.shutdown is running that script.

> > Oh, great, so I have to hit ^C on the console to get the system 3,000
> > miles away at which I typed "shutdown -r now" over the network to shut
> > down.

> If you actually tried this, you'll note that shutdown(8) runs
> /etc/rc.shutdown on the terminal you've connected from (not
> /dev/console), and has done since /etc/rc.shutdown was implemented
> (even before rc.d was incorporated).

How does this help if you don't use the 'now' flag and log off?

> To repeat, shutdown(8) runs /etc/rc.shutdown, not init(8).
> If you don't want to run /etc/rc.shutdown (and therefore run the rc.d
> scripts with tagged with "KEYWORDS: shutdown"), use `reboot' instead of
> `shutdown -r now'.

Eek! 'reboot' is a really bad move on one of my systems. I had to make
shutdown run /etc/rc.d/swap* [1]  or the mirrored swap doesn't get
unconfigured and I end up with dirty parity which takes a considerable
time to rebuild.

> On a side note: I leave you to ponder the sanity of remoting managing
> (important) boxes that you don't have the remote console of... However,
> I know it's unavoidable at times.

The box on my desk at home doesn't have something sitting on the console
It's just too much hassle for what that box is. Why should that mean I
don't to everything possible to make it remotely managable?

 [1] Yes, I am aware of the (in)sanity of this. I've looked into what it
     would take to cleanly do this in the kernel just before shutdown
     and it looked a little involed. Not to mention raising various
     issues. I haven't had time to look into this further since.
I turn my back to the wind
To catch my breath.
Before I start off again
Driven on.