From owner-freebsd-performance@FreeBSD.ORG Wed May 7 10:41:02 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 196A337B401 for ; Wed, 7 May 2003 10:41:02 -0700 (PDT) Received: from otter3.centtech.com (moat3.centtech.com [207.200.51.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id 38BF243FB1 for ; Wed, 7 May 2003 10:41:01 -0700 (PDT) (envelope-from anderson@centtech.com) Received: from centtech.com (electron.centtech.com [204.177.173.173]) by otter3.centtech.com (8.12.3/8.12.3) with ESMTP id h47Hew56061894; Wed, 7 May 2003 12:40:58 -0500 (CDT) (envelope-from anderson@centtech.com) Message-ID: <3EB94516.5070503@centtech.com> Date: Wed, 07 May 2003 12:40:38 -0500 From: Eric Anderson User-Agent: Mozilla/5.0 (X11; U; Linux i386; en-US; rv:1.0.1) Gecko/20020823 Netscape/7.0 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Clement Laforet References: <3EB67822.3070802@centtech.com> <20030505182756.093fb1c3.sheep.killer@cultdeadsheep.org> <3EB6A0BF.1040803@centtech.com> <20030506042044.GA84589@c7.campus.utcluj.ro> <3EB922B7.2080002@centtech.com> <20030507170155.GA13015@c7.campus.utcluj.ro> <20030507193247.6f60584f.sheep.killer@cultdeadsheep.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit cc: freebsd-performance@freebsd.org Subject: Re: NAT performance tweaks X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 07 May 2003 17:41:02 -0000 Clement Laforet wrote: >>Would certainly be nice to make it as a sysctl. It's a pain in the ass >>to edit it manually everytime I cvsup. :/ > > > Using a sysctl variable is a BAD idea for NAT table. If this variable is > if it can be) modified, all the NAT table must be reinitialized, because > of hash key. You should have a table size which minimizes hash > collisions, and then avoids loops and/or hypothetical race conditions. Well, why not make it a /boot/loader.conf settable sysctl, so while the system is running, it is "read only", and only settable on boot. Is there a flaw in that thinking? Eric -- ------------------------------------------------------------------ Eric Anderson Systems Administrator Centaur Technology Attitudes are contagious, is yours worth catching? ------------------------------------------------------------------