From owner-freebsd-bugs Thu May 17 4:33:11 2001 Delivered-To: freebsd-bugs@freebsd.org Received: from freebsd.dk (fw-rl0.freebsd.dk [212.242.86.114]) by hub.freebsd.org (Postfix) with ESMTP id 184D737B423 for ; Thu, 17 May 2001 04:33:07 -0700 (PDT) (envelope-from sos@freebsd.dk) Received: (from sos@localhost) by freebsd.dk (8.11.3/8.11.3) id f4HBX4457528; Thu, 17 May 2001 13:33:04 +0200 (CEST) (envelope-from sos) From: Søren Schmidt Message-Id: <200105171133.f4HBX4457528@freebsd.dk> Subject: Re: ATA driver on 4.3 is too slow In-Reply-To: <01051713264277800@Mercur.biofrontera.de> "from valgog@biofrontera.de at May 17, 2001 01:25:43 pm" To: valgog@biofrontera.de Date: Thu, 17 May 2001 13:33:04 +0200 (CEST) Cc: freebsd-bugs@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL88 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-1 Sender: owner-freebsd-bugs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org It seems valgog@biofrontera.de wrote: > Hello, > > Hier is a problem that we have with a FreeBSD 4.3-REALEASE and I > think that it is more a bug that a problem. > > We wanted to run a system with installed 3 Promise FastTrack ATA/100 > controllers with 10 70GB IDE Hard Drives. Internal IDE controllers > were turned off and a SCSI Hard Drive was used as booting device. > > On 4.2-RELEASE using vinom we managed to run RAID-10 (or what vinom > calles RAID-10) and we had perfomance of 30 MB/sec sequential writing > to this ata harddrives. But the 4.2 system was unstable and crushed 3 > times in 2 days during the havy load. > On 4.3 system it seems to be more stable (but we did run it only > several hours) and the performance was horrible: > about 8 MB/sec for writing. Most of the impact is due to FreeBSD now shipping with Write Caching off as default (see man ata for details), that will hurt even if you run with softupdates. If you dont care about possible dataloss at power failures you can enable WC and get your performance back... -Søren To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message