From owner-freebsd-arch Mon Feb 10 1:14:40 2003 Delivered-To: freebsd-arch@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2E96137B401 for ; Mon, 10 Feb 2003 01:14:39 -0800 (PST) Received: from HAL9000.homeunix.com (12-233-57-224.client.attbi.com [12.233.57.224]) by mx1.FreeBSD.org (Postfix) with ESMTP id 7F58B43FE0 for ; Mon, 10 Feb 2003 01:13:18 -0800 (PST) (envelope-from dschultz@uclink.Berkeley.EDU) Received: from HAL9000.homeunix.com (localhost [127.0.0.1]) by HAL9000.homeunix.com (8.12.6/8.12.5) with ESMTP id h1A9DHoH005418; Mon, 10 Feb 2003 01:13:17 -0800 (PST) (envelope-from dschultz@uclink.Berkeley.EDU) Received: (from das@localhost) by HAL9000.homeunix.com (8.12.6/8.12.5/Submit) id h1A9DHKE005417; Mon, 10 Feb 2003 01:13:17 -0800 (PST) (envelope-from dschultz@uclink.Berkeley.EDU) Date: Mon, 10 Feb 2003 01:13:17 -0800 From: David Schultz To: Poul-Henning Kamp Cc: arch@FreeBSD.ORG Subject: Re: Our lemming-syncer caught in the act. Message-ID: <20030210091317.GD5165@HAL9000.homeunix.com> Mail-Followup-To: Poul-Henning Kamp , arch@FreeBSD.ORG References: <31118.1044817404@critter.freebsd.dk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <31118.1044817404@critter.freebsd.dk> Sender: owner-freebsd-arch@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Thus spake Poul-Henning Kamp : > I have suspected our syncer of being subobtimal for some time, based > simply on my perception of the disk-light on my laptop and the > dynamics of the "dirty" counter in systat. > > I played with the new GEOM I/O statistics stuff and guess what: I > caught it in the act: > > http://phk.freebsd.dk/misc/disk.png > > Green is the number of unfinished requests. > > Most of the red "impulse" represents one request finishing after > as many milliseconds as it is tall. The remainder of the impulses > cover more than one request, the height is then the average of the > time it has taken to service them. > > An image is worth a thousand words, but in this case it only > says three words: "Man, that sucks!". When a large file times out, a significant amount of I/O can be generated. This is still far better than the old syncer that flushed everything every 30 seconds. The reasons for this behavior are explained in src/sys/ufs/ffs/README. After reading that, do you still think it makes sense to try to do better? To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message