From owner-freebsd-questions  Wed May 16  7: 9:28 2001
Delivered-To: freebsd-questions@freebsd.org
Received: from MPI-Softtech.Com (mpi.mpi-softtech.com [208.60.120.177])
	by hub.freebsd.org (Postfix) with ESMTP id 52ED137B422
	for <questions@FreeBSD.ORG>; Wed, 16 May 2001 07:09:23 -0700 (PDT)
	(envelope-from dleimbac@MPI-Softtech.Com)
Received: from mpi.mpi-softtech.com (mpi.mpi-softtech.com [208.60.120.177]); 
          by MPI-Softtech.Com (8.9.3/8.9.3/MPI-Softtech/evision: 1.3 $) with SMTP; 
          id JAA02571 for <questions@FreeBSD.ORG>; Wed, 16 May 2001 09:09:16 -0500 (CDT)
Message-Id: <200105161409.JAA02571@MPI-Softtech.Com>
Date: Wed, 16 May 2001 09:09:16 -0500 (CDT)
From: Dave Leimbach <dleimbac@MPI-Softtech.Com>
Reply-To: Dave Leimbach <dleimbac@MPI-Softtech.Com>
Subject: Re: python fork call raised my load over 400!
To: questions@FreeBSD.ORG
MIME-Version: 1.0
Content-Type: TEXT/plain; charset=us-ascii
Content-MD5: r0W72Uz9s40iU+fC9FS9rQ==
X-Mailer: dtmail 1.3.0 CDE Version 1.3 SunOS 5.7 sun4u sparc 
Sender: owner-freebsd-questions@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Its not a question of being a benchmark.... Its about being able to handle a 
sudden huge load and come back gracefully.

For that pupose its probably the best test I can think of.

Dave
>Date: Wed, 16 May 2001 15:04:11 +0200
>From: Michael Schuster <michael.schuster@sun.com>
>X-Accept-Language: en
>MIME-Version: 1.0
>To: dleimbac@earthlink.net
>CC: questions@FreeBSD.ORG, arch@FreeBSD.ORG
>Subject: Re: python fork call raised my load over 400!
>Content-Transfer-Encoding: 7bit
>
>dave wrote:
>> 
>> If you have a block of free time today check this out!
>> 
>> I keyed this in interactively with Python <it wasn't a file>
>> ----SNIP--------
>> 
>> import os
>> 
>> while 1:
>>         os.fork()
>> -----SNIP-------
>
>this is a classical fork bomb, and the system behaved very much as
>designed. If you're using this to compare Linux to FreeBSD, you'd better
>reconsider and get yourself proper benchmarks.
>
>btw: pls. don't cross-post, questions is quite enough.
>
>
>> This user run program brought my system to a load of 419 with the system
>> using
>> 94% of the resources and 500 user processes on my AMD Duron 800 box with
>> 256MB RAM...
>
>of course: every new process needs resources, and as new processes get more
>CPU share than older ones, the newly forked processes would immediately
>fork again.
> 
>> I don't know that the processor/RAM is relevant but I could not fork
>> anymore!
>
>of course you couldn't, you completely filled up your machine are were
>still doing so - getting a word in egdeways was impossible.
> 
>> My ultimate question is ... should I be comparing FreeBSD to Linux?
>> Does it really matter if Linux is performing better or worse than FreeBSD?
>
>see above - this about the worst type of "benchmark" I've ever seen.
>
>> Still  a user process probably shouldn't be able to hose the whole system
>> IMHO.
>
>sorry, that's the way Unix's fair-share scheduler works.
> 
>for more details, look into "Design and Implementation of 4.4 BSD"
>
>HTH
>Michael
>-- 
>Michael Schuster          / Michael.Schuster@sun.com
>Sun Microsystems GmbH     / (+49 89) 46008-2974 | x62974
>Sonnenallee 1, D-85551 Kirchheim-Heimstetten
>
>Recursion, n.: see 'Recursion'


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message