Date: Thu, 28 Oct 2004 08:19:53 +0200 From: Andreas Braukmann <braukmann@tse-online.de> To: Julian Elischer <julian@elischer.org> Cc: Kris Kennaway <kris@obsecurity.org> Subject: Re: growfs - Using on mounten FS - planned?- TWE probs.. Message-ID: <D9C8795E534EAD5BAA2C87CB@[192.168.225.210]> In-Reply-To: <41801D44.1060504@elischer.org> References: <20041027194551.54ced1e7.lehmann@ans-netz.de> <20041027175440.GA60198@xor.obsecurity.org> <20041027201443.44f3bb03.lehmann@ans-netz.de> <5521E0229FFC0CAECEA5EA17@[192.168.225.210]> <417FFE5A.2070405@elischer.org> <85C963600BB244852D738B91@[192.168.225.210]> <41801D44.1060504@elischer.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--On Mittwoch, 27. Oktober 2004 15:12 Uhr -0700 Julian Elischer <julian@elischer.org> wrote: > Andreas Braukmann wrote: >> --On Mittwoch, 27. Oktober 2004 13:00 Uhr -0700 Julian Elischer >> <julian@elischer.org> wrote: >> >> twe(4) and a bunch of big S-ATA drives in RAID-5 configuration. >> Nothing really impressive nowadays. Sorry, for the new boxes I've meant twa(4), of course. Sorry for this very misleading information. > I'm having infrequent corruption problems with 3ware 7500-8 cards . hmmm. 7500. The 72xx and 7500 series didn't fulfill my expectations concerning RAID-5 performance. I employed them only in RAID 10 confi- gurations. The 7506 series was (IMHO) the first 3Ware device delive- ring a usable RAID-5 implementation. > do you check your data? yes, regularly. > We store GB sized files and we checksum them and we check the checksums over time.... > Over 7 servers we see a couple of file drop chunks of data over a weeks worth > of writing.. either 128KB chunks are not written or the chunk corresponding to > 1 drive's part of a 128KB write never happens. Some of them are defintly firmware > problems but some of them might be driver or OS problems too (or firmware). I can't reproduce this on 7506-8 / RAID-5 / 64 kB stripe size. Just don't bother with non default stripe size on the 3Ware's; the per- formance will suck and its prone to obscure (read: not so widely ex- posed) firmware bugs. > This is with 4.8 + the driver from their website.. and the driver from 4.10 stock twe(4) from RELENG_4 (shortly after 4.8 or 4.10) and 5.1-current (end of september 2003). -Andreas
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?D9C8795E534EAD5BAA2C87CB>