Subject: re: /dev/log -> /var/run/logsocket
To: matthew green <mrg@eterna.com.au>
From: What EES eet man? <greywolf@starwolf.com>
List: tech-userlevel
Date: 01/03/1999 11:24:46
I think we should just stick with the /var/run/*.pid stuff.  Besides it
being historical MO, /var/run IS system state!.

I've got a shutdown.rc which does something to the equivalent of:

for file in /var/run/*.pid; do
        read pid < $file	# works around sendmail.pid
        kill $pid
	rm -f $file
done;

There's some other stuff in there (such as: it keeps a list of pids
so that if they didn't die, it can SIGKILL them as a last resort), but
that's basically it.  Nice and clean.

So what's the problem with pids in /var/run?  I certainly consider
pidfiles part of "system state", and a shell script to handle this
stuff is not out of the question.  Come on, folks, this is UNIX, essentially.
We can always write a killdaemon script and put it in the distribution if
need be.



				--*greywolf;
--
[ ] PEACE    [ ] FREEDOM	Pick one.