From owner-freebsd-isp Sat Jul 27 11:59:25 2002 Delivered-To: freebsd-isp@freebsd.org Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0A02B37B400 for ; Sat, 27 Jul 2002 11:59:20 -0700 (PDT) Received: from ren.sasknow.com (ren.sasknow.com [207.195.92.131]) by mx1.FreeBSD.org (Postfix) with ESMTP id 66C8C43E3B for ; Sat, 27 Jul 2002 11:59:19 -0700 (PDT) (envelope-from ryan@sasknow.com) Received: from localhost (ryan@localhost) by ren.sasknow.com (8.11.6/8.11.6) with ESMTP id g6RIxGT53281; Sat, 27 Jul 2002 12:59:16 -0600 (CST) (envelope-from ryan@sasknow.com) Date: Sat, 27 Jul 2002 12:59:16 -0600 (CST) From: Ryan Thompson To: George Ramos Cc: freebsd-isp@FreeBSD.ORG Subject: Re: Hard drives keep dying.....loooking for some advice In-Reply-To: <003c01c23583$a12d8ca0$20fea8c0@ddd> Message-ID: <20020727124224.B50718-100000@ren.sasknow.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-isp@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org George Ramos wrote to freebsd-isp@FreeBSD.ORG: > This is our main server and it is handling Web, DNS, mail, mail junk > filter, virus scanning, etc On top of that, the environment where > this server is located is rather warm...(not necessarily hot but > warm). Before moving to an air-conditioned office building, we used to have servers running continuously under load in 35C + temperatures. The worst that happened was fans dying every four or five months. We didn't take any special cooling precautions... most systems were just standard OEM builds. I'll say that I feel more comfortable in a controlled 21C environment (for the equipment as well as the staff :-). > We can make the environment cooler (at a cost) but at this point we > are not sure if the problem is mainly caused by heat or by overload > (and of course possibly a combination of both, but we are not sure > that heat is the issue because the server run fine until we added > the new services.....) Keep your equipment cool, yes, but read below. I would be very surprised if heat was (is) the only variable in this equation. > The situation seems to be "borderline", the last crash happend a few > hours after we "re-enabled" the virus scanning routine (which we had > disabled because of the previous crash.......) As a matter of fact, > the server had been running very nicely for months untill we put in > place the two new mail services (junk filter and virus scan) at > which point hard drives started to die...... I would hesitate to say that this is purely heat related. You're talking about 20GB and 40GB drives at 5.4K/7.2K RPM, so you're using IDE drives, which, in my experience, don't heat up that much. The drives would have to be getting *really* hot to fail this quickly, and with as much repeatability as you seem to be experiencing. Most of the time, heat just shortens the life of the equipment, and won't cause an abrupt failure. Check voltages. Regulate power. Check all of your *other* components (MB, controller, power supply, CPU) for signs of failure. Rule of thumb: if it burns your thumb when you touch it, it's too hot. :-) > I mean, do drives get "substantially" hotter as the work load > increases? Yes. > Can anyone make any recommendations on drives? Any particular > brand/model (IDE) that seems to tolerate heat better than others? I'm comfortable with all of the major brands. (WD, Maxtor, IBM, ...) > Do 5400 RPM drives run cooler than 7200 Yes. > Do 20 Gb H/D's run any cooler that 40 Gb's? We can use either) I don't really know. I wouldn't expect the 40GBs to run a *lot* hotter than the 20s, so the increased density might help you cool more efficiently. (For example, it's easier to cool one drive at 50C than it is to cool two drives at 45C (temperatures fictional :-)). > What else can we do to lessen the work load on these drives? Spread the load across a few machines. Analyse the performance of your applications. The drives may be working harder than necessary. Note that for some applications, SCSI can handle a higher load much more efficiently. If you make regular backups, or your data is not extremely critical, RAID 0 (striping) might improve performance (and decrease load). But, before running through hoops to optimize your system, try running it on a different system to help rule out flaky hardware. - Ryan -- Ryan Thompson SaskNow Technologies - http://www.sasknow.com 901 1st Avenue North - Saskatoon, SK - S7K 1Y4 Tel: 306-664-3600 Fax: 306-244-7037 Saskatoon Toll-Free: 877-727-5669 (877-SASKNOW) North America To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message