From owner-freebsd-fs@FreeBSD.ORG Sat Apr 8 14:39:24 2006 Return-Path: X-Original-To: freebsd-fs@freebsd.org Delivered-To: freebsd-fs@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 9018C16A404 for ; Sat, 8 Apr 2006 14:39:24 +0000 (UTC) (envelope-from user@dhp.com) Received: from shell.dhp.com (shell.dhp.com [199.245.105.1]) by mx1.FreeBSD.org (Postfix) with ESMTP id 267C743D46 for ; Sat, 8 Apr 2006 14:39:24 +0000 (GMT) (envelope-from user@dhp.com) Received: by shell.dhp.com (Postfix, from userid 896) id 20CD831341; Sat, 8 Apr 2006 10:39:22 -0400 (EDT) Date: Sat, 8 Apr 2006 10:39:22 -0400 (EDT) From: Ensel Sharon To: freebsd-fs@freebsd.org Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Subject: Re: UFS2 with 4TB disk _totally absurd_ X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 08 Apr 2006 14:39:24 -0000 On Thu, 6 Apr 2006, Scott Long wrote: > Ensel Sharon wrote: > >>The FDISK and bsdlabel schemes simply cannot deal with >2TB. You'll > >>need to either put your filesystem directly on the storage device > >>without and slices/labels, or use GPT to create logical partitions. > > > > > > 2TB filesystems are _not large_. FreeBSD should expect 2-4TB filesystems > > to be in common use in peoples _living rooms_, never mind in the office or > > datacenter. > > > > So 5.x was a total wash in terms of UFS2 and snapshots, largefiles, etc., > > 6.0 still doesn't have working filesystem quotas or snapshots, and it > > seems, doesn't support modern (circa 2004) hard drives. > > > > Maybe a little less time working on FreeBSD 23.0 ... ? > > > > What are you talking about? UFS2, the filesystem, supports storage > volumes up to 2^63 blocks in size, and filesystems themselves of > more than 2^53 blocks in size. There is no 2TB limit in UFS2, and I've > personally created filesystems that are indeed much larger than that.. > These sizes were supported in 2004, and they are supported in 2006. > What is limited is the FDISK and BSDLABEL formats, which were designed > in the early 80's to handle up to 2^32 blocks. Neither of these prevent > you from creating a large filesystem. Maybe you're looking to have a > single large volume to hold both your boot filesystem and your data > filesystem? That's generally a bad idea since it puts more things into > the path of a failure. Try doing what most people do, which is to boot > off of a 2 disk mirror (go big and get 500GB disks if you want) and have > your data on a separate array that is more redundant and doesn't need to > use the above partition formats. > > Alternatively, find a PC that understands how to boot off of GPT > partitions, and use that format. It's not FreeBSD's fault that the PC > BIOS uses the FDISK format. Go complain to IBM and Microsoft for not > having the foresight to future-proof their partition format 25 years > ago. I'm not saying that you can't create 2+ TB filesystems. I am saying that there are all sorts of problems with them, from snapshotting them, to fsck, to easy creation with sysinstall. And all the while, the reaction has been a non-chalant dismissiveness, presumably based on the fact that 2+ TB filesystems are out of the norm, or are "frontier" hardware or whatever. I am writing to inform you that that is not the case, and has not been for quite some time. I know many people with 2+ TB filesystems in their houses - I have one _attached to my television_. So whereas the fact that fsck and snapshots and dump and restore and df and quotas are all broken on 2+ TB: http://www.freebsd.org/projects/bigdisk/index.html The thought was that that would get cleaned up circa 5.2 or something. Here we are at 6.1 and it's still broken. Perhaps when commodity disks exceed 50% of the size of the max known-working FreeBSD partition the urgency will increase ?