From owner-freebsd-stable@FreeBSD.ORG Sun Nov 7 05:57:07 2004 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2396116A4CE for ; Sun, 7 Nov 2004 05:57:07 +0000 (GMT) Received: from gw.catspoiler.org (217-ip-163.nccn.net [209.79.217.163]) by mx1.FreeBSD.org (Postfix) with ESMTP id D900B43D1D for ; Sun, 7 Nov 2004 05:57:06 +0000 (GMT) (envelope-from truckman@FreeBSD.org) Received: from FreeBSD.org (mousie.catspoiler.org [192.168.101.2]) by gw.catspoiler.org (8.13.1/8.13.1) with ESMTP id iA75ux5W050510; Sat, 6 Nov 2004 21:57:03 -0800 (PST) (envelope-from truckman@FreeBSD.org) Message-Id: <200411070557.iA75ux5W050510@gw.catspoiler.org> Date: Sat, 6 Nov 2004 21:56:59 -0800 (PST) From: Don Lewis To: scrappy@hub.org In-Reply-To: <20041107005645.O46679@ganymede.hub.org> MIME-Version: 1.0 Content-Type: TEXT/plain; charset=us-ascii cc: freebsd-stable@FreeBSD.org Subject: Re: Don's changes to fsck on 4.x ... X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 07 Nov 2004 05:57:07 -0000 On 7 Nov, Marc G. Fournier wrote: > On Sat, 6 Nov 2004, Don Lewis wrote: > >> On 6 Nov, Marc G. Fournier wrote: >>> >>> Well, finally had a reason to use it, and its running right now ... seems >>> a bit slower in phase 2 then before ... is to be expected? Looking >>> through the patch, it seems that all pass's were affected, so this might >>> be now the norm ... after ~39minutes running on a very large file system, >>> hitting ctl-T periodically, I'm up to about 50% through Phase 2 ... so far >>> *knock on wood* no errors being generated by fsck itself, but that doesn't >>> mean anything :) >> >> Under normal circumstances, there shouldn't be any noticeable difference >> in performance. If there are a lot of zero link count files, phase 1 >> should be very slightly faster because the zero link count file list no >> longer needs to be allocated, and phase 4 should be a lot faster. Most >> of the time in phases 1 and 2 is consumed by disk reads. The only >> change to phase 2 was the addition of the new inode states to a couple >> of case statements and an if statement which should not affect the >> amount of I/O done and CPU time would only be affected by a miniscule >> amount, so I would not expect any change to the performance of that >> phase. > > Hindsight is 20/20, but I should have trap' the output ... I saw several > 'ZERO LENGTH DIRECTORY' messages in Phase 4 still ... should I have seen > any at all? Yes. The behaviour should be exactly the same as before. The only difference is that phase 4 should run a lot faster if there are a lot of UNREF files and directories.