From owner-freebsd-questions@FreeBSD.ORG Sat Apr 29 05:36:55 2006 Return-Path: X-Original-To: freebsd-questions@freebsd.org Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5301F16A400 for ; Sat, 29 Apr 2006 05:36:55 +0000 (UTC) (envelope-from ijeff@sandbox.ca) Received: from sandbox.ca (zaphod.sandbox.ca [199.246.50.125]) by mx1.FreeBSD.org (Postfix) with ESMTP id 7AC5143D45 for ; Sat, 29 Apr 2006 05:36:54 +0000 (GMT) (envelope-from ijeff@sandbox.ca) Received: from [127.0.0.1] (sandbox.ca [199.246.50.125]) by sandbox.ca (8.12.6/8.12.6) with ESMTP id k3T5aox1027205 for ; Sat, 29 Apr 2006 01:36:51 -0400 (EDT) (envelope-from ijeff@sandbox.ca) Mime-Version: 1.0 (Apple Message framework v749.3) To: freebsd-questions@freebsd.org Message-Id: <97E10AD3-097B-408F-9B48-9C06BE2116E8@sandbox.ca> From: Ian Jefferson Date: Sat, 29 Apr 2006 14:36:50 +0900 X-Mailer: Apple Mail (2.749.3) X-Spam-Status: No, score=-1.2 required=5.0 tests=ALL_TRUSTED,HTML_FONT_BIG, HTML_MESSAGE autolearn=failed version=3.1.0-igj_main_2005_08_07 X-Spam-Checker-Version: SpamAssassin 3.1.0-igj_main_2005_08_07 (2005-09-13) on zaphod.sandbox.ca Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Content-Transfer-Encoding: 7bit X-Content-Filtered-By: Mailman/MimeDel 2.1.5 Subject: gvinum help under 6.0 release X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 29 Apr 2006 05:36:55 -0000 Hi folks, Well I think I'm out of ideas in my experience with gvinum. I need some help. I cannot get gvinum to work for me at all in setting up a raid5 set. This is the first FreeBSD gizmo that I've run into that has proven dangerously unreliable. Each time I use it I get a panic, and one of these tries kept the machine from booting until I did a bsdlabel -B /dev/adxx for each of three drives. (it just did it again) I was surprised that even a dd if=/dev/zero of=/dev/adxx would not clean up things. I read somewhere that geom does something preventing overwrite of parts of the device. I'm hoping someone will point out something I'm doing horribly wrong. Synopsis: The drives in question are ad4, ad8 and ad10 all identical disks. One attempt under amd64 6.0 release with this gvinum patch I get panics with i386 6.1-RC1 http://wikitest.freebsd.org/moin.cgi/GvinumMoveRename bsd2# cat vinum_r5.config drive a device /dev/ad4s1d drive b device /dev/ad8s1d drive c device /dev/ad10s1d volume r5vol plex org raid5 15g sd length 5g drive a sd length 5g drive b sd length 5g drive c bsd2# cat sdisk.bsdlabel # for vinum configuraiton # good for /dev/ad4s1 /dev/ad8s1 and /dev/ad10s1 # # # /dev/ad4s1: 8 partitions: # size offset fstype [fsize bsize bps/cpg] a: 390721889 16 unused 0 0 c: 390721905 0 unused 0 0 # "raw" part, don't edit d: 50g 17 vinum bsd2# cat clean.sh dd if=/dev/zero of=/dev/ad4 bs=1k count=100 fdisk -I ad4 bsdlabel -R /dev/ad4s1 sdisk.bsdlabel dd if=/dev/zero of=/dev/ad8 bs=1k count=100 fdisk -I ad8 bsdlabel -R /dev/ad8s1 sdisk.bsdlabel dd if=/dev/zero of=/dev/ad10 bs=1k count=100 fdisk -I ad10 bsdlabel -R /dev/ad10s1 sdisk.bsdlabel The story so far: I recently purchased a Gigabyte GA-8VT880P (VIA PT880 Pro Chipset) and put an Intel Celeron D 336 (Intel EMB64T) processor in it. I suppose by today's standards this is a pretty low end board but it's way fast for what I need. On this system the MB was replacing an older AMD Athalon board. I'm using this system to study an upgrade path to a 4.7 system I'm running. I stupidly decided to install the 6.0-RELEASE amd64. At this point I'm still running the Generic kernel. I say stupidly because the target system is really going to run an i386-RELEASE of 5.x or 6.x. Later I repented, after gvinum frustration, re-sliced and moved my original install onto slice 2. For disk I have to PATA drives that I have been using for some time and 3 Samsung SATA drives. All but one of the drives is attached to the MB controller. I have an addon "Buffalo" SATA/PATA card (also ~cheap) with a VT6421L chipset in it. I'll comment that one of the SATA connections on the addon board does not seem to function correctly with the amd release. I have not tested it yet with my i386 RC-1 yet. HOWEVER I did run some default iozone tests on all the working SATA and PATA drives so I'm fairly confident that this the working SATA connection really works OK. I also have a SATA "backplane". These are called various things but the basic idea is to put three 3.5" drives in two 5.25" HH external slots. This one has "hot swap" capability (dubious IMHO). For me this is a convenient mechanical place to put disks. Again I'm confident that this is OK since I ran the iozone tests with the drives in this enclosure. What I am trying to do: What I'm studying is how to put together a "software" raid5 volume and generally I like what I was reading about vinum as a flexible tool to manage space. I'm not terribly concerned with IO performance since I'm still calibrated to 5MB/s sustained throughput and keep wondering why 2kb of code keeps getting repacked into 35MB's of bloat. (see "grumble") above :-). TIA