From owner-freebsd-scsi@FreeBSD.ORG Wed May 12 12:05:39 2004 Return-Path: Delivered-To: freebsd-scsi@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 117A616A4CE; Wed, 12 May 2004 12:05:39 -0700 (PDT) Received: from bremen.shuttle.de (bremen.shuttle.de [194.95.249.251]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5E9D643D46; Wed, 12 May 2004 12:05:38 -0700 (PDT) (envelope-from schweikh@schweikhardt.net) Received: by bremen.shuttle.de (Postfix, from userid 10) id 56E803BBB2; Wed, 12 May 2004 21:05:37 +0200 (CEST) Received: from hal9000.schweikhardt.net (localhost [127.0.0.1]) i4CJ4Te3001372; Wed, 12 May 2004 21:04:29 +0200 (CEST) (envelope-from schweikh@hal9000.schweikhardt.net) Received: (from schweikh@localhost)i4CJ4TVQ001371; Wed, 12 May 2004 21:04:29 +0200 (CEST) (envelope-from schweikh) Date: Wed, 12 May 2004 21:04:29 +0200 From: Jens Schweikhardt To: Scott Long Message-ID: <20040512190429.GA1315@schweikhardt.net> References: <40A26356.409@freebsd.org> <20040512183231.GB3137@schweikhardt.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20040512183231.GB3137@schweikhardt.net> User-Agent: Mutt/1.5.6i cc: freebsd-scsi@freebsd.org Subject: Re: 2120S Stripe - abysmal performance X-BeenThere: freebsd-scsi@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: SCSI subsystem List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 12 May 2004 19:05:39 -0000 On Wed, May 12, 2004 at 08:32:31PM +0200, Jens Schweikhardt wrote: # On Wed, May 12, 2004 at 11:48:06AM -0600, Scott Long wrote: # ... # # I can't remember how the 2100 BIOS works, I'll have to take a look. # # I thought that it was an attribute that was set when the array is # # created. # # Yes, and once its set at array creation time it apparently can not be # changed. I deleted the array and created it anew to disable read # caching. This did not yield any perfomance gain by itself. After I # changed the cables I now have 79MB/s (formerly only 29MB/s, a single # disk has a sustained read speed of 71MB/s at the outer sectors). Next # thing will be going back from forced U160 mode to U320. I'll report the # results, so watch this space :-) Went back to U320 and enabled packetized--even with the Adaptec cable that came with the controller no joy :-( At first, dd reported 83MB/s but after skipping the first 1G the controller(?) made a high pitched beep, and this was in dmesg aac0: **Monitor** SCSI bus reset issued on channel 0 aac0: **Monitor** <...repeats 2 more times> aac0: **Monitor** Drive 0:14:0 online on container 0: aac0: **Monitor** Drive 0:15:0 online on container 0: aac0: **Monitor** SCSI bus reset issued on channel 0 aac0: **Monitor** <...repeats 1 more times> aac0: **Monitor** Drive 0:14:0 online on container 0: aac0: **Monitor** Drive 0:15:0 online on container 0: aac0: **Monitor** SCSI bus reset issued on channel 0 aac0: **Monitor** Drive 0:15:0 returning error aac0: **Monitor** ID(0:15:0) - Cmd[0x28] failed (retries exhausted) aac0: **Monitor** Alarm Adapter Event -- turning Alarm On! aac0: **Monitor** Container 0 returning error on drive 0:15:0 aacd0: hard error cmd=read 2551359-2551486 aacd0: hard error cmd=read 5013567-5013694 ... aacd0: hard error cmd=read 35094591-35094718 aacd0: hard error cmd=read 35094719-35094846 aacd0: hard error cmd=read 35094591-35094718 aacd0: hard error cmd=read 35094719-35094846 aacd0: hard error cmd=read 37601343-37601470 aacd0: hard error cmd=read 37601471-37601598 aacd0: hard error cmd=read 37601343-37601470 aacd0: hard error cmd=read 37601471-37601598 aacd0: hard error cmd=read 37601343-37601470 aacd0: hard error cmd=read 37601471-37601598 aac0: **Monitor** ID(0:15:0) - Cmd[0x2a] failed (retries exhausted) aac0: **Monitor** Last Adapter Alarm Event -- Alarm turning off! Error probing container 0 aacd0: detached aac0: **Monitor** Drive 0:14:0 online Hmm. Should I still go for another U320 cable? Regards, Jens -- Jens Schweikhardt http://www.schweikhardt.net/ SIGSIG -- signature too long (core dumped)