From owner-freebsd-questions@FreeBSD.ORG Wed May 5 15:47:17 2010 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 44E8B1065673 for ; Wed, 5 May 2010 15:47:17 +0000 (UTC) (envelope-from kpielorz_lst@tdx.co.uk) Received: from mail.tdx.com (mail.tdx.com [62.13.128.18]) by mx1.freebsd.org (Postfix) with ESMTP id DBE518FC25 for ; Wed, 5 May 2010 15:47:16 +0000 (UTC) Received: from HPQuadro64.dmpriest.net.uk (HPQuadro64.dmpriest.net.uk [62.13.130.30]) (authenticated bits=0) by mail.tdx.com (8.14.3/8.14.3/Kp) with ESMTP id o45FXtwb084071 (version=TLSv1/SSLv3 cipher=DHE-DSS-AES256-SHA bits=256 verify=NO) for ; Wed, 5 May 2010 16:33:55 +0100 (BST) Date: Wed, 05 May 2010 16:33:34 +0100 From: Karl Pielorz To: freebsd-questions@freebsd.org Message-ID: X-Mailer: Mulberry/4.0.8 (Win32) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline Subject: Is it safe to increase / double kern.maxvnodes under FreeBSD 6.4 i386? X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 05 May 2010 15:47:17 -0000 Hi, We had some issues at the weekend that left one of our machines with a very, very large sendmail queue... While we were trying to sort it out we noticed the machine takes over 4 minutes to go through the queue (i.e. 'mailq'). I noticed the machine was hovering around the maxvnode limit - so I upped it. The 'sweet spot' appears to be: sysctl -w kern.maxvnodes=250000 That cuts the time to run a mailq from over four minutes down to 12 seconds. The machine has 2Gb of RAM, and is 'moderately' loaded (normally) - is it wise to leave that setting at 250,000 - or is it likely to cause other issues (i.e. kernel memory issues) - is there any metric I can look at / check to see if we can get away with leaving it that high? The only stuff I can seem to find on the 'net mostly concerns upping it under amd64 as that uses a different mapping method for vnodes - I can't seem to find anything that covers increasing it that much (2.5 * the default of 100,000) under i386. As it obviously makes a huge difference for us, I'd love to leave it in place - but don't want to risk anything drastic like a panic. Thanks, -Karl