From owner-freebsd-net@FreeBSD.ORG Mon Oct 3 10:45:52 2005 Return-Path: X-Original-To: net@FreeBSD.org Delivered-To: freebsd-net@FreeBSD.ORG Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id CBDBE16A41F for ; Mon, 3 Oct 2005 10:45:52 +0000 (GMT) (envelope-from glebius@FreeBSD.org) Received: from cell.sick.ru (cell.sick.ru [217.72.144.68]) by mx1.FreeBSD.org (Postfix) with ESMTP id 2A36843D45 for ; Mon, 3 Oct 2005 10:45:51 +0000 (GMT) (envelope-from glebius@FreeBSD.org) Received: from cell.sick.ru (glebius@localhost [127.0.0.1]) by cell.sick.ru (8.13.3/8.13.3) with ESMTP id j93AjnLQ071927 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Mon, 3 Oct 2005 14:45:49 +0400 (MSD) (envelope-from glebius@FreeBSD.org) Received: (from glebius@localhost) by cell.sick.ru (8.13.3/8.13.1/Submit) id j93AjmOc071926; Mon, 3 Oct 2005 14:45:48 +0400 (MSD) (envelope-from glebius@FreeBSD.org) X-Authentication-Warning: cell.sick.ru: glebius set sender to glebius@FreeBSD.org using -f Date: Mon, 3 Oct 2005 14:45:48 +0400 From: Gleb Smirnoff To: Ferdinand Goldmann Message-ID: <20051003104548.GB70355@cell.sick.ru> References: <4341089F.7010504@jku.at> Mime-Version: 1.0 Content-Type: text/plain; charset=koi8-r Content-Disposition: inline In-Reply-To: <4341089F.7010504@jku.at> User-Agent: Mutt/1.5.6i Cc: net@FreeBSD.org Subject: Re: dummynet, em driver, device polling issues :-(( X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 03 Oct 2005 10:45:52 -0000 On Mon, Oct 03, 2005 at 12:31:59PM +0200, Ferdinand Goldmann wrote: F> Initial observations showed that the system had a very high interrupt load, F> most of the times > 50%, and lagged a little bit. Plus, performance-testing F> with tcpspray yielded bad results at around 10MB/s, or even worse. So we F> tried enabling device polling on the em0 interface. After some testing, we F> set HZ=1000 and kern.polling.burst_max: 300. This made things slightly F> better, our tcpspray tests now reaching a ~13MB/s throughput. F> F> However, now the input errors on the em0 interface were rapidly rising: F> Name Mtu Network Address Ipkts Ierrs Opkts Oerrs F> Coll F> em0 1500 00:0e:0c:6d:4c:f9 3738461290 8605788 3613603979 F> 0 0 F> F> So for now, the polling is disabled again. All I can say: I have faced this problem, too. :( This is not problem in polling, but in em. -- Totus tuus, Glebius. GLEBIUS-RIPN GLEB-RIPE