From owner-freebsd-scsi@FreeBSD.ORG Mon Jul 13 10:05:03 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 176921065674 for ; Mon, 13 Jul 2009 10:05:03 +0000 (UTC) (envelope-from freebsd-scsi@m.gmane.org) Received: from ciao.gmane.org (main.gmane.org [80.91.229.2]) by mx1.freebsd.org (Postfix) with ESMTP id 9BD158FC1A for ; Mon, 13 Jul 2009 10:05:02 +0000 (UTC) (envelope-from freebsd-scsi@m.gmane.org) Received: from list by ciao.gmane.org with local (Exim 4.43) id 1MQI9E-0002VZ-Vf for freebsd-scsi@freebsd.org; Mon, 13 Jul 2009 09:48:36 +0000 Received: from lara.cc.fer.hr ([161.53.72.113]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Mon, 13 Jul 2009 09:48:36 +0000 Received: from ivoras by lara.cc.fer.hr with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Mon, 13 Jul 2009 09:48:36 +0000 X-Injected-Via-Gmane: http://gmane.org/ To: freebsd-scsi@freebsd.org From: Ivan Voras Date: Mon, 13 Jul 2009 11:48:24 +0200 Lines: 65 Message-ID: References: <7f834c1c0907070405mdab72u3223d90b84b2a201@mail.gmail.com> <20090707173629.GA33007@in-addr.com> <7f834c1c0907102141n5d55f786keb55f44126cd94c2@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@ger.gmane.org X-Gmane-NNTP-Posting-Host: lara.cc.fer.hr User-Agent: Thunderbird 2.0.0.21 (X11/20090615) In-Reply-To: <7f834c1c0907102141n5d55f786keb55f44126cd94c2@mail.gmail.com> Sender: news Subject: Re: iscsi_initiator 2.2.2 slow to write and read X-BeenThere: freebsd-scsi@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: SCSI subsystem List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 13 Jul 2009 10:05:03 -0000 Jiansong Liu wrote: > On Wed, Jul 8, 2009 at 1:36 AM, Gary Palmer wrote: > >> On Tue, Jul 07, 2009 at 07:05:03PM +0800, Jiansong Liu wrote: >>> Hello All, >>> >>> I downloaded the latest iscsi_initiator 2.2.2 from Danny's FTP, >>> and compile kernel and iscontrol with 7.2 STABLE GENERIC config, only got >>> about 800KB/Sec speed to a DELL MD3000i disk array, I searched in >>> this malling list and found a patch, but it seems is not for this version >>> (2.2.2), any advice would be great appreciated. >>> >>> here are some details about my system: >>> >>> # sysctl net.iscsi >>> net.iscsi.driver_version: 2.2.2 >>> net.iscsi.isid: ???DIB00 >>> net.iscsi.sessions: 1 >>> net.iscsi.0.targetname: >>> iqn.1984-05.com.dell:powervault.md3000i.60024e80004e71a9000000004a011120 >>> net.iscsi.0.targeaddress: 192.168.4.182 >>> net.iscsi.0.stats: recv=594 sent=594 flags=0x0000039f pdus-alloc=16 >>> pdus-max=16 cws=426 cmd=157 exp=157 max=300 stat=159 itt=157 >>> net.iscsi.0.douio: 0 >>> net.iscsi.0.pid: 754 >>> >>> # cat /etc/iscsi.conf >>> scpstorage01 { >>> initiatorname =iqn.2005-01.il.ac.huji.cs::static01.com >>> TargetName >>> >> =iqn.1984-05.com.dell:powervault.md3000i.60024e80004e71a9000000004a011120 >>> TargetAddress =192.168.4.182:3260,1 >>> tags =4 >>> } >> I'm no expert, but if that "tags" line refers to SCSI tags then that >> number is too small. Try increasing it and see if that helps to >> address the problem. >> >> Regards, >> >> Gary >> > > Hi Gary, > > Thanks for the advice, I have tried set tags to 32 and 64, > but it could not get more performance on IO (90MB write and 56MB read). > I think the writing is good enough for my application, but the reading still > is a problem. > > # dd if=/home/mogdata/test.foo of=/dev/null > 2048000+0 records in > 2048000+0 records out > 1048576000 bytes transferred in 17.670549 secs (59340318 bytes/sec) Are you really reading sector by sector? Try adding bs=1m to the dd line. There is something else you can try to increase sequential reading, if that's what you're interested in: increase the vfs.read_max sysctl to 32 or 64. On the other hand, it will do nothing to improve, and may harm random read performance. Test with your actual application before drawing conclusions.