From owner-cvs-all Tue Dec 19 17:16:38 2000 From owner-cvs-all@FreeBSD.ORG Tue Dec 19 17:16:34 2000 Return-Path: Delivered-To: cvs-all@freebsd.org Received: from relay.nuxi.com (nuxi.cs.ucdavis.edu [169.237.7.38]) by hub.freebsd.org (Postfix) with ESMTP id 002F237B400; Tue, 19 Dec 2000 17:16:32 -0800 (PST) Received: from dragon.nuxi.com (Ipittythefoolthattrustsident@trang.nuxi.com [209.152.133.57]) by relay.nuxi.com (8.9.3/8.9.3) with ESMTP id RAA10748; Tue, 19 Dec 2000 17:16:32 -0800 (PST) (envelope-from obrien@NUXI.com) Received: (from obrien@localhost) by dragon.nuxi.com (8.11.1/8.11.1) id eBK1GVC85079; Tue, 19 Dec 2000 17:16:31 -0800 (PST) (envelope-from obrien) Date: Tue, 19 Dec 2000 17:16:30 -0800 From: "David O'Brien" To: Warner Losh Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sbin/newfs newfs.8 Message-ID: <20001219171630.A84990@dragon.nuxi.com> Reply-To: obrien@FreeBSD.org References: <200012192155.eBJLt8n75030@freefall.freebsd.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <200012192155.eBJLt8n75030@freefall.freebsd.org>; from imp@FreeBSD.org on Tue, Dec 19, 2000 at 01:55:08PM -0800 X-Operating-System: FreeBSD 5.0-CURRENT Organization: The NUXI BSD group X-Pgp-Rsa-Fingerprint: B7 4D 3E E9 11 39 5F A3 90 76 5D 69 58 D9 98 7A X-Pgp-Rsa-Keyid: 1024/34F9F9D5 Sender: obrien@NUXI.com Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Tue, Dec 19, 2000 at 01:55:08PM -0800, Warner Losh wrote: > Log: > o Add an example for a large file system. > Not Objected to by: hackers, doc I can't find this on hackers, so I have to ask directly. PHK suggested `newfs -b 16384 -f 4096 -c 100' a while back. Rod Grimes reported: I seemed to have run into some problems with some stuff that did not like these and ended up deploying at the standard 8192/1024 -u 4096 due to them, so not a lot of testing occured. I think a few commits have been made since then to fix larger than 8K block file systems, so perhaps a more complete test with bug reports is due.... Peter reported: Incidently, I found that using a frag to block ration != 8 was a certain killer for a news filesystem. ie: 8k/1k worked beautifully, and so did 4k/ 0.5k. If I did 4k/1k or 8k/2k, the system would badly trash the file system by corrupting the block allocation bitmaps within 2 or 3 days of running at full load. I was never able to track this down though as it was only occurring on a live system that I wasn't allowed to go back to 4k:1k for debugging once it was working. The 4:1 ratio of 16k/4k rings alarm bells to me.. My question is how tested are these new officially suggested values, and how does the above comments apply today. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message