From owner-freebsd-current Tue Oct 6 16:27:16 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id QAA26720 for freebsd-current-outgoing; Tue, 6 Oct 1998 16:27:16 -0700 (PDT) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from implode.root.com (implode.root.com [198.145.90.17]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id QAA26667 for ; Tue, 6 Oct 1998 16:26:57 -0700 (PDT) (envelope-from root@implode.root.com) Received: from implode.root.com (localhost [127.0.0.1]) by implode.root.com (8.8.5/8.8.5) with ESMTP id QAA00703; Tue, 6 Oct 1998 16:26:05 -0700 (PDT) Message-Id: <199810062326.QAA00703@implode.root.com> To: Terry Lambert cc: enkhyl@hayseed.net, mishania@demos.net, current@FreeBSD.ORG Subject: Re: -current panics.. In-reply-to: Your message of "Tue, 06 Oct 1998 19:57:15 -0000." <199810061957.MAA18722@usr08.primenet.com> From: David Greenman Reply-To: dg@root.com Date: Tue, 06 Oct 1998 16:26:05 -0700 Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG >> >This same kind of situation can occur with a network-intensive app running >> >under Solaris, too. From my own experience with tuning OSs (specifically >> >Solaris 2.x), I'd agree with David's assessment. Not having delved that >> >far into the details of FreeBSD "leaks" as he probably has, I can't say >> >with certainty that there are none, but I'm inclined to believe him. :-) >> >> I should say here that I'm not saying that there aren't any "leak" bugs, >> either, just that I have looked into this several time and found nothing but >> mis-tuned systems. I'm also a bit incredulous since wcarchive does far more >> TCP/IP than most other machines in the world, has months of uptime and no >> detectable buffer leaks. I want to be clear about this: If there was a bug >> that I could identify, I'd fix it immediately. "Show me the mon^H^H^Hbug!" > >I kind of have a question in this area... > >If it's possible to tune the system's static resource allocations, >then why not make the allocations dynamic so that they are self >tuning, and dampen the hysteresis loop so that you don't get wild >swings? > > >It's always seemed odd to me that there would ever be an "out of >mbufs!" message when there was, in fact, room available to allocate >more mbufs... The kernel address space and layout is a carefully crafted and tuned finite resource. There are several cases where decent performance demands that certain things be statically allocated, and virtual memory for network buffers is one of those things. -DG David Greenman Co-founder/Principal Architect, The FreeBSD Project To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message