From owner-freebsd-threads@FreeBSD.ORG Wed Mar 24 11:35:51 2004 Return-Path: Delivered-To: freebsd-threads@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id EDD2616A4CE for ; Wed, 24 Mar 2004 11:35:51 -0800 (PST) Received: from mail.asn.net (mail.asn.net [66.235.231.4]) by mx1.FreeBSD.org (Postfix) with SMTP id B276B43D2F for ; Wed, 24 Mar 2004 11:35:51 -0800 (PST) (envelope-from kris-fbsd@asn.net) Received: (qmail 13821 invoked by uid 80); 24 Mar 2004 19:35:46 -0000 Received: from 68.228.211.254 (SquirrelMail authenticated user kgale) by mail.asn.net with HTTP; Wed, 24 Mar 2004 12:35:46 -0700 (MST) Message-ID: <50726.68.228.211.254.1080156946.squirrel@mail.asn.net> In-Reply-To: References: <51632.68.14.252.60.1079987672.squirrel@mail.asn.net> Date: Wed, 24 Mar 2004 12:35:46 -0700 (MST) From: "Kris Gale" To: "Daniel Eischen" User-Agent: SquirrelMail/1.4.2 MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 Importance: Normal cc: freebsd-threads@freebsd.org Subject: Re: Loaded MySQL 4.0.18 w/ KSE running nicely X-BeenThere: freebsd-threads@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Threading on FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 24 Mar 2004 19:35:52 -0000 > And I use still use your script to test MySQL, except that > I use don't override the default for key_buffer, and I also > patch MySQL to use process scope threads. I just tested out the patch for mysql, and everything seems to be running very smoothly. Not only was I able to get 1800 threads (90 * 20) running (which was a huge stumbling block for me earlier), but I'm looking over at 4000 threads that managed to connect (200 * 20). Before trying the patch, I synced to -current. There was some talk about some recent scheduler changes maybe helping things. This didn't seem to help at all. Before I patched mysql, I was getting frequent lockups with as few as 100 threads. Since the key_buffer value is tuned for the amount of RAM on my test system, I've left it in place... I'm going to continue by trying to test actual queries I'd see in production to see how this compares to my 4-stable system. Kris