Date: Thu, 23 Aug 2001 13:44:37 -0700 (PDT) From: Hodge Podge <nicole@unixgirl.com> To: Lawrence Farr <freebsd-hw@sixforty.co.uk> Cc: Marc Rassbach <marc@milestonerdl.com>, freebsd-hardware@freebsd.org, Michael VanLoon <MichaelV@edifecs.com>, Darryl Okahata <darrylo@soco.agilent.com> Subject: RE: 3ware stuff not ready for heavy duty useage-followup Message-ID: <XFMail.010823134437.nicole@unixgirl.com> In-Reply-To: <004101c12bc3$81e74900$c80aa8c0@lfarr>
next in thread | previous in thread | raw e-mail | index | archive | help
Very Nice. Thanks! Well, I'm not sure what the power supply na-sayers might want as a test, but do you have a way to force data into the drives from an outside source? I was doing this for abt 6-8 Hrs when mine failed :( Lots of small files. Who knows.. maybe that was enough to woogle the power supply, maybe its a Via Chipset interaction problem... The bugger is what to do abt it? I talked it over with my Boss and I want to see if we can set something up with 3ware to do some testing. We house hundredes of thousands of small (~300k) files) and maybe that is the problem. Maybe its bad drives.. Altho we got them from different places. But damn, process of elimination for something like this is no small thing. Nicole On 23-Aug-01 Lawrence Farr wrote: > I left my server (previous email below) running a benchmark that I made > up that took > a 1mb file and cat'ed to another file until it got to 1Gb. > I then read the file back by cat'ing to /dev/null, and repeated until I > filled the > partition (140Gb). It did this 3 times in total and got results like: > > Pass 23 - 1048576 kb written in 115 seconds, at 9118 kb/Sec > Pass 23 - 1048576 kb read in 15 seconds, at 69905 kb/Sec > Pass 24 - 1048576 kb written in 116 seconds, at 9039 kb/Sec > Pass 24 - 1048576 kb read in 15 seconds, at 69905 kb/Sec > > It all works fine, but the writes are awful. Iozone seems to agree with > me: > > File size set to 10240 KB > Time Resolution = 0.000001 seconds. > Processor cache size set to 1024 Kbytes. > Processor cache line size set to 32 bytes. > File stride size set to 17 * record size. > random random > > KB reclen write rewrite read reread read write read > 10240 4 9397 9461 266043 266458 236609 188083 253962 > > Anyone got any ideas for tuning this? Or is that it performance wise? > How about splitting the drives to 4 stripe sets, and Raid 5 across them > with vinum? > > Anyone else want to see any specific tests? > >>-----Original Message----- >>From: Lawrence Farr [mailto:freebsd-hw@sixforty.co.uk] >>Sent: 22 August 2001 14:59 >>To: 'Hodge Podge' >>Cc: 'mike.wentz@3ware.com'; 'freebsd-hardware@FreeBSD.ORG'; >>'Borja Marcos'; 'Michael VanLoon'; 'val@picturetrail.com' >>Subject: RE: 3ware stuff not ready for heavy duty useage- ATTN 3ware >> >> >>Im just formatting an 8 drive (Maxtor 536DX 100Gb) 7000 series >>array, 400W supply. >> >>Pick a benchmark and I'll run it for a while. > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-hardware" in the body of the message ******* |\ __ /| (`\ ******* * * | o_o |__ ) ) * * * * // \\ * * * Nicole Harrington | AKA Hodge Podge * ----------------------(((---(((-------------------------------- nicole@unixgirl.com http://www.unixgirl.com/ webmistress@dangermouse.org http://www.dangermouse.org/ nicole@deviantimages.com http://www.deviantimages.com/ -- Powered By Coca-Cola and FreeBSD -- "The world is run by those who show up" -- The Best Place for Your Web Site - www.WebWeaver.net -- ---------------------------------------------------------------- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.010823134437.nicole>