From owner-freebsd-stable Mon Nov 8 12:36:45 1999 Delivered-To: freebsd-stable@freebsd.org Received: from dialup422.apex.dp.ua (dialup422.apex.dp.ua [195.24.155.122]) by hub.freebsd.org (Postfix) with ESMTP id 7A8D914EF8 for ; Mon, 8 Nov 1999 12:36:22 -0800 (PST) (envelope-from voland@dialup422.apex.dp.ua) Received: (from voland@localhost) by dialup422.apex.dp.ua (8.9.3/8.9.3) id WAA70143; Mon, 8 Nov 1999 22:35:19 +0200 (EET) (envelope-from voland) Mime-version: 1.0 Content-type: text/plain; charset="koi8-r" Content-transfer-encoding: 8bit X-Comment-To: Yuri Vorobyev To: Yuri Vorobyev Cc: freebsd-stable@FreeBSD.ORG Subject: Re: ccd performance degradation & reboot problem References: <4497.991108@yamalinfo.ru> From: Vadim Belman In-Reply-To: Yuri Vorobyev's message of "Mon, 8 Nov 1999 11:56:34 +0500" Date: 08 Nov 1999 22:34:38 +0200 Message-ID: <85zowoiuxt.fsf@dialup422.apex.dp.ua> Lines: 69 X-Mailer: Gnus v5.7/Emacs 20.4 Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Hi Yuri! On 08 Nov 99 at 08:56, "Yuri" (Yuri Vorobyev) wrote: Yuri> I have made some tests with bonnie. Yuri> 3.1-RELEASE -------Sequential Output-------- ---Sequential Input-- Yuri> --Random-- -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- Yuri> --Seeks--- Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU Yuri> K/sec %CPU /sec %CPU 100 10085 83.9 13683 38.7 4808 15.5 7716 75.0 Yuri> 14429 26.2 213.7 4.4 Yuri> FreeBSD 3.3-STABLE #0: Mon Nov 8 Yuri> -------Sequential Output-------- ---Sequential Input-- Yuri> --Random-- -Per Char- --Block--- -Rewrite-- -Per Char- --Block--- Yuri> --Seeks--- Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU Yuri> K/sec %CPU /sec %CPU 100 1866 16.1 1802 4.3 1828 6.1 8177 80.1 14384 Yuri> 25.7 213.8 5.2 Yuri> Kernel config was the same. Yuri> Write speed has fallen very much!! This may not be related to the problem's cause, but I usually try different interleave factors using a script which records bonnie's output for later analysis. Differences are significant for various interleaves, but never falls below actual drive speed. So, this rather looks like a wrong way but may be useful information in future. (BTW, my ccd uses interleave factor of 256 and used to be 512 in previous configuration. Both values based on benchmark results.) You may also make sure that you didn't check your da0 drive second time. 8) There are few perfomance issues about IBM DCAS drives. Yuri> syncing disks... done (da1:ahc0:0:1:0) SYNCHRONIZE CACHE: CDB: 35 0 Yuri> 0 0 0 0 0 0 0 0 (da1:ahc0:0:1:0) ILLEGAL REQUEST asc:20,0 Yuri> (da1:ahc0:0:1:0) invalid command operation code sks:c0,0 Yuri> Rebooting... Yuri> or Yuri> #reboot Yuri> syncing disks... done (da1:ahc0:0:1:0) SYNCHRONIZE CACHE: CDB: 35 0 Yuri> 0 0 0 0 0 0 0 0 (da1:ahc0:0:1:0) error code 9 Rebooting... Yuri> Error code change randomly (?) Yuri> And freeze... Have you had a look on server's console during benchmarking? I'd expect some SCSI-related error messages on it. You may suffer either from hardware problems or from a driver bug. The second possibility seems to be rather improbable. But the first one is practically definite. Check your cable connection, run SCSI BIOS hardware test. I.e. make sure that da1 works properly. Or not... Yuri> System have ADAPTEC 2940 ULTRA / ULTRA W BIOS v1.25 Ahm, I read some recomendations about upgrading SCSI BIOS to any of 1.3x version. Actually it's been done for my board. -- /Voland Vadim Belman E-mail: voland@plab.ku.dk To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message