Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753440AbYLFUzn (ORCPT ); Sat, 6 Dec 2008 15:55:43 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752751AbYLFUzc (ORCPT ); Sat, 6 Dec 2008 15:55:32 -0500 Received: from an-out-0708.google.com ([209.85.132.245]:27141 "EHLO an-out-0708.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752733AbYLFUzb (ORCPT ); Sat, 6 Dec 2008 15:55:31 -0500 Message-ID: Date: Sat, 6 Dec 2008 15:55:29 -0500 From: "Kyle Moffett" To: "Justin Piszcz" Subject: Re: Have the velociraptors in a test system now, checkout the errors. Cc: "Michael Tokarev" , linux-kernel@vger.kernel.org, linux-raid@vger.kernel.org, xfs@oss.sgi.com, smartmontools-support@lists.sourceforge.net In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Content-Disposition: inline References: <493A5E62.1020508@msgid.tls.msk.ru> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1677 Lines: 31 On Sat, Dec 6, 2008 at 7:11 AM, Justin Piszcz wrote: > As far the PSU, just btw/FYI, Velociraptors consume ~4-5 watts a piece, my > entire system used ~100-120watts with all 12 velociraptors on a 650 watt PSU > (now moved into a test system). You should actually break it down further than that. During system development recently, the company I work for was using some 350W PSUs with funny connectors patched in place of the standard molex to drive the 5V power for some small embedded system testbeds and their drives. We were hard-power-cycling the systems by unplugging them from the bus and we kept having problems with the *other* embedded board resetting when we did so. Turns out the "350W" PSU was only rated to supply 100W to the 5V leads, and our 55W systems (40W for board and 15W for drive) were a little too far past the edge. In addition, most hard drives, motherboards, etc have circuits powered separately from the 3.3V, 5V, and 12V busses. If your *load* is balanced too heavily on one or the other of the supplies, you can see extremely weird problems. When the 12V-powered spindle and spindle-controller on your HDD loses power but the 5V-powered SATA or SAS interface does not, its internal state machine gets all kinds of messed up. If you can get ahold of one, I'd recommend finding an oscilliscope to hook up to the 5V and 12V lines into those drives. Cheers, Kyle Moffett -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/