From owner-freebsd-fs@FreeBSD.ORG Wed Jul 9 13:34:17 2008 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 17F2F1065670 for ; Wed, 9 Jul 2008 13:34:17 +0000 (UTC) (envelope-from lopez.on.the.lists@yellowspace.net) Received: from mail.yellowspace.net (mail.yellowspace.net [80.190.200.164]) by mx1.freebsd.org (Postfix) with ESMTP id A51508FC1A for ; Wed, 9 Jul 2008 13:34:16 +0000 (UTC) (envelope-from lopez.on.the.lists@yellowspace.net) Received: from five.intranet ([88.217.71.135]) (AUTH: LOGIN lopez.on.the.lists@yellowspace.net) by mail.yellowspace.net with esmtp; Wed, 09 Jul 2008 15:34:14 +0200 id 00330909.000000004874BE56.00008961 Message-Id: <743AA903-03CB-4F43-A30B-9D06C58A4EAC@yellowspace.net> From: Lorenzo Perone To: mike In-Reply-To: Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit Mime-Version: 1.0 (Apple Message framework v924) Date: Wed, 9 Jul 2008 15:34:14 +0200 References: X-Mailer: Apple Mail (2.924) Cc: freebsd-fs@freebsd.org Subject: Re: Thinking of using ZFS/FBSD for a backup system 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: Wed, 09 Jul 2008 13:34:17 -0000 On 08.07.2008, at 08:15, mike wrote: > Just wanted to sanity check here before investing some money and time > into this solution... > > Also if anyone wants to reply to me off list with hardware that works > well for FBSD 7 + ZFS I'd be grateful :) On 09.07.2008, at 13:17, Kris Kennaway wrote: > ZFS is what it is, several of us have shown that it is possible to > tune memory parameters to make it fit into a FreeBSD kernel, and > users can either take that for what it's worth, or decide that ZFS > is not for them. > > Kris I know I'm definitively NOT going to make any friends here this way ;) BUT: At the present time, my impression is that if you need/want to put anything business-critical on a ZFS pool: go opensolaris. I got to know it on a SUN box, at least the bit to run and compile a few things, and it was definitively a pain compared to the good old FreeBSD hier(7). But for as far as zfs goes, I can sleep at night. Not that I wouldn't prefer the other way around, I'm one of those whom others have to stop before he starts installing FreeBSD on everything that has a CPU and runs as a server ;) I'm sure zfs on FreeBSD will be at least as stable (and, easilly, even better performing) as soon as pjb has the time to share his newest patches. I'll be one of the first csup'ping on RELENG_7 as soon as it's done. It's a way too sexy filesystem not to do so, and that's the reason why this kind of threads pop up regularly... As for my current experience, I have a box that backs up two offices every night and snapshots several filesystems, which is now running stably a zfs pool since about 2 months (which is almost a record for that box). It's a 7.0-STABLE #8 Sat Apr 26 10:10:53 CEST 2008, amd64 with 2 GB of RAM and the following in /boot/loader.conf: vm.kmem_size=900M vm.kmem_size_max=900M vfs.zfs.arc_max=300M vfs.zfs.prefetch_disable=1 However I'm happy that it is down the corridor, so if the whole thing gets stuck again, or panics, I can still watch it while it reboots on the console... BTW I can't recommend to rely on ufs snapshots while waiting for zfs to become stable - I tried and it almost destroyed that filesystem after a few runs (besides, it takes ages to finish a snapshot, compared to zfs). Regards, (and long live FreeBSD, it's half my life atm) Lorenzo