From owner-freebsd-current Mon Nov 9 09:10:32 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id JAA25075 for freebsd-current-outgoing; Mon, 9 Nov 1998 09:10:32 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from whistle.com (s205m131.whistle.com [207.76.205.131]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id JAA25061 for ; Mon, 9 Nov 1998 09:10:30 -0800 (PST) (envelope-from archie@whistle.com) Received: (from smap@localhost) by whistle.com (8.7.5/8.6.12) id JAA17543 for ; Mon, 9 Nov 1998 09:10:14 -0800 (PST) Received: from bubba.whistle.com(207.76.205.7) by whistle.com via smap (V1.3) id sma017541; Mon Nov 9 09:10:09 1998 Received: (from archie@localhost) by bubba.whistle.com (8.8.7/8.6.12) id JAA01978 for current@freebsd.org; Mon, 9 Nov 1998 09:10:09 -0800 (PST) From: Archie Cobbs Message-Id: <199811091710.JAA01978@bubba.whistle.com> Subject: Re: The infamous dying daemons bug In-Reply-To: <199811091626.IAA21393@hub.freebsd.org> from "garman@earthling.net" at "Nov 9, 98 11:28:08 am" To: current@FreeBSD.ORG Date: Mon, 9 Nov 1998 09:10:09 -0800 (PST) X-Mailer: ELM [version 2.4ME+ PL38 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG garman@earthling.net writes: > > Totally unrelated to the problem. It seems, so far as I was able to > > characterize, to happen to daemons which are *swapped out* at the time > > of the memory shortage. If it's active enough to still be in core, it > > doesn't get spammed. > > > Yes, this fits the symptoms i'm seeing here with samba. The daemon > will be fine as long as it has not been swapped out; after that, it's > *poof*. That's also probably why people with heavy-use samba servers > haven't seen this problem. Vague observations.. - samba, inetd, sendmail all do a lot of forking (which may be nothing other than a common need for more memory) - samba uses memory mapping to implement file locking I'd like to try to confirm/deny that memory mapping is one required ingredient of the recipie. The only other known ingredient seems to be running out of swap. Could someone who is seeing this happen often recompile their kernel with memory mapping disabled, and see if that changes anything (other than making programs that use mmap() stop working)? Ie, comment out options SYSVSHM options SYSVSEM options SYSVMSG (this *does* disable mmap(), right? If not, instead patch the mmap() syscall to always return an error) -Archie ___________________________________________________________________________ Archie Cobbs * Whistle Communications, Inc. * http://www.whistle.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message