From owner-freebsd-hackers Sat Oct 26 07:33:46 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id HAA29106 for hackers-outgoing; Sat, 26 Oct 1996 07:33:46 -0700 (PDT) Received: from mailgate.execpc.com (root@mailgate.execpc.com [169.207.16.5]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id HAA29100 for ; Sat, 26 Oct 1996 07:33:43 -0700 (PDT) Received: from earth.execpc.com (jgreco@earth [169.207.16.1]) by mailgate.execpc.com (8.7.6/8.7.5) with ESMTP id JAA07358; Sat, 26 Oct 1996 09:17:41 -0500 Received: (from jgreco@localhost) by earth.execpc.com (8.7.6/8.7) id JAA10206; Sat, 26 Oct 1996 09:17:14 -0500 (CDT) From: Joe Greco Message-Id: <199610261417.JAA10206@earth.execpc.com> Subject: HELP! :-( Hitting datasize limit To: hackers@freebsd.org Date: Sat, 26 Oct 1996 09:17:14 -0500 (CDT) Reply-To: jgreco@ns.sol.net X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk HELP! I have a very very large application: Usenet news (innd). I have been in the habit of starting it "unlimit"'ed for quite some time now. It starts off executing at about 55MB and will generally grow to double that within 12 hours. However, lately, it has been dying multiple times daily with malloc errors (using phkmalloc, but given the apparent cause I do not think this is the problem). The current situation would tend to have caused it to grow >> 110MB... and when I did a little looking into the problem, malloc was indeed returning NULL, which caused me to scratch my head until I typed "unlimit; limit"... cputime unlimited filesize unlimited datasize 131072 kbytes stacksize 65536 kbytes coredumpsize unlimited memoryuse unlimited descriptors 4136 memorylocked 251412 kbytes maxproc 2067 S***! 128MB absolute datasize limit on a process?!?! This machine has 256MB RAM and if INN wants to use it all, I really do not care. I can probably somehow hack a "raise" to this limit myself, but I have no idea what nasty side effects might be waiting for me and I am wondering what the idea or architectural reasons behind this "absolute" limit is. What is recommended in this case? ... JG