* Craig McLean ([EMAIL PROTECTED]) wrote:
> Ok so the rc script explains why not all of the S/K entries get run on
> runlevel changes.
>
> My next question: The init man page says all of the processes in the same
> process group as init that aren't also in the next run level should get a
> SIGTERM and a SIGKILL signal killing them. This doesn't seem to be
> happening. cron and syslog are both in the same process group as init but
> they aren't getting killed. As an experiment I removed the cron and syslog
> entries from rc4.d and then trasferred to run level 4 from 2. Yet cron and
> syslog weren't killed.
>
> Incidentally I don't think the removing of entries from the run level
> directories should make a difference anyways because I know that init
> doesn't actually have any knowledge of rc directories.
>
> I do know that when I transfer between various user defined run levels the
> signals aren't getting sent, but when I transfer between from a user defined
> run level to a system runlevel the signals do get sent.
true -- init knows nothing of the /etc/rc?.d/ directories.
but init knows about /etc/inittab, which contains the line
l0:0:wait:/etc/init.d/rc 0
l1:1:wait:/etc/init.d/rc 1
l2:2:wait:/etc/init.d/rc 2
l3:3:wait:/etc/init.d/rc 3
l4:4:wait:/etc/init.d/rc 4
l5:5:wait:/etc/init.d/rc 5
l6:6:wait:/etc/init.d/rc 6
# Normally not reached, but fallthrough in case of emergency.
z6:6:respawn:/sbin/sulogin
which reveal that, on entry to, say, runlevel 3, the script
"/etc/init.d/rc" is run with a sole argument "3".
looking at the /etc/init.d/rc script, it uses its sole argument
to determine which /etc/rc?.d/ directory to look in; there, it
runs any K* scripts with an argument "stop", and then any S*
scripts with an argument "start".
if a K* script doesn't send any "kill -HUP `pidof mydaemon`" then
the process "mydaemon", if running, will stay running. likewise,
if a S* script doesn't actually launch a daemon, then no daemon
is launched.
in february of 2000 (when the manpage for init was last tweaked,
from what i can tell) init may have operated they way it's
describe there. but the way it works on debian potato is that it
calls a script (/etc/init.d/rc) that calls scripts
(/etc/rc*.d/[KS]*) with certain arguments. they're spozed to be
the ones to do the job -- and this is a nice, new, modular
layout.
you've discovered the biggest achilles' heel of the *nix world:
"the documentation was probably right at one point, but..."
to determine that "point", check the "last-modified" date and
keep exploring if you need to...
--
DEBIAN NEWBIE TIP #39 from Roy Culley <[EMAIL PROTECTED]>
:
Wondering why the pundits say that CSH SCRIPTS ARE CONSIDERED
HARMFUL? Although "tcsh" has improved on "csh" there are still
issues. For the full scoop, read
http://www.faqs.org/faqs/unix-faq/shell/csh-whynot/
Also see http://newbieDoc.sourceForge.net/ ...
