From owner-freebsd-stable@FreeBSD.ORG Sun Jul 16 02:30:16 2006 Return-Path: X-Original-To: freebsd-stable@freebsd.org 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 3B38916A4DF for ; Sun, 16 Jul 2006 02:30:16 +0000 (UTC) (envelope-from javier@skyhawk.kjsl.com) Received: from skyhawk.kjsl.com (skyhawk.kjsl.com [69.36.241.138]) by mx1.FreeBSD.org (Postfix) with ESMTP id DA24D43D46 for ; Sun, 16 Jul 2006 02:30:15 +0000 (GMT) (envelope-from javier@skyhawk.kjsl.com) Received: by skyhawk.kjsl.com (Postfix, from userid 1001) id 38A87B818A; Sat, 15 Jul 2006 22:30:15 -0400 (EDT) Date: Sat, 15 Jul 2006 22:30:15 -0400 From: Javier Henderson To: Matthew Seaman Message-ID: <20060716023014.GE86544@skyhawk.kjsl.com> References: <20060714165735.L15214@bunning.skiltech.com> <44B8094C.3040007@rogers.com> <20060715011112.GB32358@skyhawk.kjsl.com> <44B88B51.10607@infracaninophile.co.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <44B88B51.10607@infracaninophile.co.uk> User-Agent: Mutt/1.4.2.1i Cc: Mike Jakubik , freebsd-stable@freebsd.org Subject: Re: Intel ICH7R RAID controller working on 6.1/STABLE? X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 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, 16 Jul 2006 02:30:16 -0000 * Matthew Seaman [060715 02:30]: > Javier Henderson wrote: > > * Mike Jakubik [060714 17:15]: > > >> The chipset is supported, but i wouldn't recommend onboard raid for any > >> production server. Get a real raid controller, or use gmirror if you > >> plan to mirror. I use several of these board sin production with gmirror. > > > Why do you recommend against on-board RAID controllers? > > Think about what happens if one of your disks dies. Sure, the machine will > carry on running. With an on-board controller there are two problems: > > i) How do you get notified that a disk has died > ii) How do you replace the drive > > (i) you'ld likely only find out about at reboot time, or by noticing a > change in the pattern of blinken-lights on the machine. (Don't laugh -- > it happens) Good point. The Intel motherboard I have with on-board RAID controller doesn't have a notification features as I've seen on stand-alone controllers. > (ii) is not just about having to power off the machine and swap out the > hardware: it's not uncommon for on-board RAID-1 setups to be unable to > rebuild a mirror by duplicating the good disk onto the replacement one. That > means blowing everything away and recovering from backup. By which time > you've had so much downtime that you might as well not have bothered with > RAID in the first place. Well, in my case, I mentioned I have the Intel D945PVS motherboard. Before storing valuable data, I did take out a drive (out of four in a RAID 5 configuration) while reading and writing to/from the array, and it just kept on going. Then I put the disk back, and things got slow while parity was rebuilt, but in the end the array was back to healthy status. > The advantage of a good RAID controller -- like one of the 3ware cards > -- or of gmirror is that combined with hot-swap disk (and pretty much all > SATA drives nowadays have hot-swap capability; you just need to find a > chassis with the right sort of drive bays) then you can take out the dead > disk, replace it with a good one and rebuild the array *without taking the > machine down*. > > gmirror will alert you to failures in the nightly e-mail if you enable > the 406.status-gmirror periodic script. Similarly a good hardware RAID > controller will have a system level control application to let you interface > with the card from the OS level, and it will have some mechanism for alerting > the admin to problems. Yes, there are indeed good advantages to stand-alone controllers, and in some cases they justify the expense. Thanks for taking the time to post a reply. -jav