From owner-freebsd-current@FreeBSD.ORG Fri Sep 3 21:01:45 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id C73BF16A4CE for ; Fri, 3 Sep 2004 21:01:45 +0000 (GMT) Received: from ganymede.hub.org (blk-222-46-91.eastlink.ca [24.222.46.91]) by mx1.FreeBSD.org (Postfix) with ESMTP id 590A743D39 for ; Fri, 3 Sep 2004 21:01:45 +0000 (GMT) (envelope-from scrappy@hub.org) Received: by ganymede.hub.org (Postfix, from userid 1000) id AC392359FE; Fri, 3 Sep 2004 18:01:43 -0300 (ADT) Received: from localhost (localhost [127.0.0.1]) by ganymede.hub.org (Postfix) with ESMTP id AB3A33538A for ; Fri, 3 Sep 2004 18:01:43 -0300 (ADT) Date: Fri, 3 Sep 2004 18:01:43 -0300 (ADT) From: "Marc G. Fournier" To: freebsd-current@freebsd.org Message-ID: <20040903175434.A812@ganymede.hub.org> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Subject: what is fsck's "slowdown"? X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Sep 2004 21:01:45 -0000 I'm running another fsck on a 4.x system ... I know 5.x brings in the background mode for it, which would definitely make my life easier, *but* ... when running fsck, it says its using up 99% of the CPU: # ps aux | grep fsck root 67 99.0 5.0 184252 184284 p0 R+ 12:46PM 254:16.68 fsck -y /vm now, its a dual CPU system ... on an MP system, is it not possible to have it parallelize on a file system, so that it makes use of all available CPU? For instance, right now, its in Phase 4 ... on a file system where ctl-t shows: load: 0.99 cmd: fsck 67 [running] 15192.26u 142.30s 99% 184284k /dev/da0s1h: phase 4: cyl group 408 of 866 (47%) wouldn't it be possible, on a dual CPU system, to have group 434 and above run on one process, while group 433 and below running on the second, in parallel? Its not like the drives are being beat up: # iostat 5 tty da0 pass0 cpu tin tout KB/t tps MB/s KB/t tps MB/s us ni sy in id 0 44 0.00 0 0.00 0.00 0 0.00 40 0 0 0 60 0 11 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 16.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 45 16.00 0 0.01 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 0.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 6 16.00 0 0.00 0.00 0 0.00 50 0 0 0 50 0 4600 16.00 3 0.04 0.00 0 0.00 49 0 1 0 50 0 18 16.00 0 0.01 0.00 0 0.00 50 0 0 0 50 0 16 16.00 1 0.01 0.00 0 0.00 50 0 0 0 50 So, it looks to me like the process is CPU bound, not disk ... Or, does 5.x's fsck already make better use of available CPUs? ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664