Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!seismo!mcvax!ukc!man.psy!neil
From: neil@man.psy.UUCP (Neil Todd at UK.AC.MAN.CS.UX)
Newsgroups: net.unix,net.unix-wizards
Subject: Re: Killing processes
Message-ID: <328@man.psy.UUCP>
Date: Wed, 5-Feb-86 08:39:20 EST
Article-I.D.: man.328
Posted: Wed Feb  5 08:39:20 1986
Date-Received: Fri, 7-Feb-86 21:49:59 EST
References: <324@man.psy.UUCP> <369@mcgill-vision.UUCP>
Reply-To: neil@man.cs.ux.UUCP (Neil Todd at UK.AC.MAN.CS.UX)
Organization: DEPT COMPUTER SCIENCE, UNIV MANCHESTER, ENGLAND
Lines: 35
Xref: watmath net.unix:7042 net.unix-wizards:16685


Thanks to everybody who replied to my query about killing processes.

There seem to be a number of schools of thought on this matter :-

1) Send a STOP to the process, then kill it at your leisure.

2) If you're using >= 4.2, renice the offending user and processes
   down to 20, this gives you more time to catch processes that
   are busy forking.

The trouble with these two is that you've still got to do a "ps",
which was the thing that was bugging me in the first place -
especially if the system is slow.

3) Use something along the lines of the zap program given in
   "The Unix Programming Environment", this does the "ps" for
   for and asks you process by process whether you want to kill
   them off.

4) Introduce a new system call that can "kill-by-uid". This is
   nice and quick, but will kill off EVERY process belonging
   to that uid.

I've opted for a combination of 3 and 4. The new system call went
in with very little effort, thanks to Ken Lalonde at University of
Waterloo for the basic code for 4.2.


Neil Todd

JANET:-	neil@uk.ac.man.cs.ux
UUCP :-	...!mcvax!ukc!man.cs.ux!neil
ARPA :- neil%uk.ac.man.cs.ux@ucl.cs.arpa   
[ Please use the ARPA route rather than UUCP if you can]