From owner-freebsd-questions Sun Feb 10 6:26:11 2002 Delivered-To: freebsd-questions@freebsd.org Received: from guru.mired.org (dsl-64-192-6-133.telocity.com [64.192.6.133]) by hub.freebsd.org (Postfix) with SMTP id B5FF937B417 for ; Sun, 10 Feb 2002 06:26:07 -0800 (PST) Received: (qmail 22404 invoked by uid 100); 10 Feb 2002 14:26:06 -0000 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <15462.33533.684435.312295@guru.mired.org> Date: Sun, 10 Feb 2002 08:26:05 -0600 To: BSD Freak Cc: questions@freebsd.org Subject: Re: What I've gathered about the soft updates discussion In-Reply-To: <16857937@toto.iv> X-Mailer: VM 6.90 under 21.1 (patch 14) "Cuyahoga Valley" XEmacs Lucid X-face: "5Mnwy%?j>IIV\)A=):rjWL~NB2aH[}Yq8Z=u~vJ`"(,&SiLvbbz2W`;h9L,Yg`+vb1>RG% *h+%X^n0EZd>TM8_IB;a8F?(Fb"lw'IgCoyM.[Lg#r\ From: "Mike Meyer" X-Delivery-Agent: TMDA/0.44 (Python 2.2; freebsd-4.5-STABLE-i386) Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG BSD Freak types: > Hiya all, > > Please excuse me and if I sound ignorant but I am really trying to get > my head around soft updates and UNIX (especially FreeBSD) file systems > in general. I think the discussion on the topic so far has been very > good and for everyone's sake (and mine) here are a few things I've > gathered so far: > > 1. Soft updates is a Good Thing(TM). According to the tuning manpage it > gauruntees file system consistency but along with write caching (enabled > in FreeBSD 4.4+ by default) can lead to data loss. At least to me this > seemed contradictory at first but I think this is what it means (and > please correct me if I am mistaken as I have not actually run tests to > prove any this this yet): if you have soft updates enabled and your > system is shut down uncleanly (ie power failure or crash) an fsck will > be run upon reboot but there will be no metadata or data corruption ( > the fsck is only to retreive lost space), however you may lose the > actual data written within the last minute or so. Your last statement is true of softupdates with drives that are operating properly. If the drives aren't operating properly, all bets are off. One way to insure that the drives operate properly is to disable write caching. This tends to slow IDE drives down considerably. The real problem here is that there is a high degree of variability in how well IDE drives follow the spec. If the drive follows the spec properly, all should be well and good. But not all of them do, and in extreme cases will tell you data is on the disk and then delay actually writing it indefinitely. So when you enable write caching on an IDE drive, you don't know how much risk you are exposing your data to. Personally, I run my IDE drives with write caching enabled and soft updates - that doesn't really add much danger once you turn on write caching - and make sure that the critical data is backed up on SCSI drives. http://www.mired.org/home/mwm/ Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message