kill_minus_9

This is part of The Pile, a partial archive of some open source mailing lists and newsgroups.



Subject: Re: Why is everyone recommending kill -9?
From: Cameron Simpson <cameron@research.canon.com.au>
Date: Thu, 14 Oct 1999 14:02:52 +1000


On Sun, Oct 10, 1999 at 08:10:58PM -0700, Todd A. Jacobs wrote:
| I see a *lot* of traffic relating to the kill command, and note that a lot
| of people are steering the newbies towards kill -9. This is a *bad* habit
| to get into.

Yep. It's because they know no better.

| The whole point of kill -9 (the SIGKILL signal) is to prevent a program
| that has run amok from trapping the signal. SIGKILL prevents programs from
| cleaning up after themselves, and really should only be used as a last
| resort.
| 
| Try 'kill <process>' first, which sends the SIGTERM signal.

Yep.

| If that
| doesn't work, try 'kill -QUIT <process>' which is slightly more forceful,
| but still gives a program one last chance to die cleanly.

No.

SIGQUIT is to get a core dump. Most programs should _not_ trap it, because you
want to know _exactly_ where it was when you nailed it.

Programs which catch SIGQUIT are BROKEN (or security critical - a minority).
It's a debugging aid. If an ordinary program catches QUIT and tidies
up, discard it:-)

Use SIGHUP (though anyone with the foresight to catch SIGHUP will probably
also catch SIGTERM).

| Then, and only
| then, should 'kill -KILL <process>' be sent.

Indeed.

===

Subject: Re: Why is everyone recommending kill -9?
From: "Todd A. Jacobs" <tajacobs@nvbell.net>
Date: Wed, 13 Oct 1999 22:31:08 -0700 (PDT)


On Thu, 14 Oct 1999, Cameron Simpson wrote:

> Programs which catch SIGQUIT are BROKEN (or security critical - a
> minority). It's a debugging aid. If an ordinary program catches QUIT
> and tidies up, discard it:-)

Heh. That would include Netscape, then. I often get race conditions in
Netscape that will respond to a SIGQUIT, but not a SIGTERM. 

===


the rest of The Pile (a partial mailing list archive)

doom@kzsu.stanford.edu