From owner-freebsd-scsi@FreeBSD.ORG Mon Jun 1 03:36:36 2009 Return-Path: Delivered-To: freebsd-scsi@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 31C701065673 for ; Mon, 1 Jun 2009 03:36:36 +0000 (UTC) (envelope-from cch@kmu.edu.tw) Received: from cc.kmu.edu.tw (cc.kmu.edu.tw [163.15.154.1]) by mx1.freebsd.org (Postfix) with ESMTP id CB9DF8FC1B for ; Mon, 1 Jun 2009 03:36:35 +0000 (UTC) (envelope-from cch@kmu.edu.tw) Received: from c199.cc.kmu.edu.tw (c199.cc.kmu.edu.tw [163.15.154.199]) by cc.kmu.edu.tw (8.14.2/8.14.2) with ESMTP id n513a0hg087547; Mon, 1 Jun 2009 11:36:01 +0800 (CST) (envelope-from cch@kmu.edu.tw) Message-ID: <4A234CA0.9010009@kmu.edu.tw> Date: Mon, 01 Jun 2009 11:36:00 +0800 From: =?UTF-8?B?IuisneW/l+aYjChDaGloLUNoYW5nIEhzaWVoKSI=?= Organization: KMU User-Agent: Thunderbird 2.0.0.21 (X11/20090422) MIME-Version: 1.0 To: Andrew Snow References: <4A23376D.4080300@kmu.edu.tw> <4A234343.7000809@modulus.org> In-Reply-To: <4A234343.7000809@modulus.org> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-KMU-MailScanner-Information: Please contact the ISP for more information X-KMU-MailScanner: Not scanned: please contact your Internet E-Mail Service Provider for details X-KMU-MailScanner-From: cch@kmu.edu.tw X-Spam-Status: No Cc: freebsd-scsi@freebsd.org Subject: Re: Some problems about iSCSI initiator 2.1.0 X-BeenThere: freebsd-scsi@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list Reply-To: cch@kmu.edu.tw List-Id: SCSI subsystem List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 01 Jun 2009 03:36:36 -0000 Dear Andrew, Thank you for your advice. We have re-done the tests by dd if=/dev/zero of=${iSCSI_TARGET_DIR}/bigf bs=1024k count=1000 But it does not make big differences. The iSCSI target da2 is still much steady than da0 and da1. The new results are at http://free.cc.kmu.edu.tw/iscsi/iotest.txt Original results with /dev/random are now at http://free.cc.kmu.edu.tw/iscsi/iotest-random.txt Andrew Snow wrote: >> Our tests are dd if=/dev/random of=${iSCSI_TARGET_DIR}/bigf bs=1024k >> count=1000 > > Try using /dev/zero instead of /dev/random, which causes excessive CPU > load on the initiator which affects the results. > > > - Andrew