From owner-freebsd-current@FreeBSD.ORG Sat Feb 18 04:11:38 2006 Return-Path: X-Original-To: freebsd-current@freebsd.org 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 B27E816A422 for ; Sat, 18 Feb 2006 04:11:38 +0000 (GMT) (envelope-from brian@aljex.com) Received: from s1tank.virtdom.com (s1tank.virtdom.com [216.240.101.50]) by mx1.FreeBSD.org (Postfix) with SMTP id 25B7743D48 for ; Sat, 18 Feb 2006 04:11:37 +0000 (GMT) (envelope-from brian@aljex.com) Received: (qmail 42570 invoked by uid 89); 18 Feb 2006 04:51:11 -0000 Received: from ool-43552092.dyn.optonline.net (HELO venti) (brian@aljex.com@67.85.32.146) by s1tank.virtdom.com with SMTP; 18 Feb 2006 04:51:11 -0000 Message-ID: <00d901c63441$684811b0$6c00000a@venti> From: "Brian K. White" To: References: <43F647CD.5060709@nikiforov.ru> Date: Fri, 17 Feb 2006 23:11:08 -0500 Organization: Aljex Software MIME-Version: 1.0 Content-Type: text/plain; format=flowed; charset="iso-8859-1"; reply-type=response Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2900.2670 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2670 Subject: Re: LSI Trouble X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 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: Sat, 18 Feb 2006 04:11:38 -0000 ----- Original Message ----- From: "Anton Nikiforov" To: Sent: Friday, February 17, 2006 5:01 PM Subject: LSI Trouble > Dear All, > sorry for sending my request to so many maillists and looks like it is OFF > TOPIC, but i have defenetley big trouble and kindly asking you to help. > > I have had server running FreeBSD-5.4p9 and it was samba file server for > MS domain (about 600 people). > > Today we were changing power supply so we shut all servers down at 17:00. > > After changing of UPS we start everything up and my LSI controller starts > to claim that disk and NVRAM configuration mismatch. > > When i tried to enter configuration console i have found out that all > disks in my array are in READY state, that means they are not a part of > any drive. > > After rebooting (via simple exitting the configuration console withiut > changing anything), server stopped to claim that configuration mismutch, > but shows 0 Logical drives configured (while before i have had 2 drives): > Raid1 for the OS (2*140GB drives) > Raid5 for the data (3*140GB drives) > And one 140GB drive as a hot spare disk > LSI FW version is G119 from Aug 27 2004 > > I'm sure that disks are still containing the information, but i do not > know how to restore the LSI data on disks to boot properly or mount this > drive somewhere on different system to get the data. > > ICP controllers support nondistructive build to build array from disks > that already contain ICP information. But i have found nothing about the > same function in LSI. And i'm afraid that this will erase (now i do not > trust LSI) my info. > > I did contact LSI support but 3 people answerred that one is on the > vacation, the other on the business trip and the third one is ill..... > nice support. THere is nothing to say :) > > In case there is no way to restore my configuration and continue to use > this server as it was could you please tell me the way to restore data > from that drives? > > Best regards, > Anton NIkiforov There is an option in the lsi cards, which unfortunately it may be too late for you to benefit from. I'll say what I do know and hopefully it will give yu something to look into in case you still get no help. Before you got that first error, there is an option you can set to tell the card what to do when there is a discrepency between the nvram and the on-disk data. 1) ask the user 2) trust the disk 3) trust the nvram When you were first seeing the error, I think you also had a chance to say "trust the nvram" or "trust the disks" It's not clear to me if you got that prompt, or if you did, how you answered it. Maybe the nvram got zapped and you hit enter and that caused it to use the (blank) nvram, thus wiping the disks (not all of them just the raid data) Maybe the disks still hold good metadata and you can get the card to read the disks. You can go int the card now and find that option (under adapter settings, possibly advanced), set it to trust the disks, and see if it comes up after a reboot. That is basically "safe" to try. It shouldn't hurt the disks whether it works or not. If not, then the raid data is gone from the disks. Then you can try this if you have 5 spares of the same disks: pull out all your live disks and put them to the side. Keep track of their positions put in 5 spares that you don't care about erasing. configure the exact same raid arrays on the new disks as on the old ones tell the card to trust the nvram instead of the disks power down put the old disks back in power up, on booting up, let it resolve the discrepency by trusting the nvram. That is pretty risky. I don't know that it works. If you had 4 spare disks (which you would need for above anyways), or better , 6, you could test it first though, by defining a simple 2 disk raid0 or 3 disk raid5 array (just not raid1, you need striping and ideally parity too, since one of your dead arrays is raid5), do a quick minimal install just to create a working booting filesystem, then deliberately zap the array information and use the other 2 (or 3) disks to create the array in nvram again, tell the card to trust the nvram, swap the first 2 (or 3) disks back in and see if they come back alive. If they do then it should be ok to use spare disks to define the original arrays into the cards nvram, tell the card to trust the nvram, then put your original real disks back in. If you don't have enough spare disks, but are desperate enough that it's worth buying some IF you knew for a fact that it world work, then I can perform the simple 2-drive & 2-drive test and let you know if it worked. I happen to have 4 old 18 gig drives and a 320-1 card and a spare server. However, I'm hitting the road for Boskone (sf convention in Boston this weekend) so I wouldn't be able to test until Monday, perhaps Monday night. You'd almost certainly have to be able to get the exact same model and size of disks too, not approximations. Hopefully LSI gets back to you with some more practical method. You know the real answer of course is, shame on you, where are your tape backups that make this nothing more than a small inconvenience while you watch a tape run for a couple hours? Get an LTO-2 drive and a supertar (BackupEDGE) and a two-week rotation of tapes and never sweat again. Brian K. White -- brian@aljex.com -- http://www.aljex.com/bkw/ +++++[>+++[>+++++>+++++++<<-]<-]>>+.>.+++++.+++++++.-.[>+<---]>++. filePro BBx Linux SCO FreeBSD #callahans Satriani Filk!