From owner-freebsd-questions@FreeBSD.ORG Mon May 24 05:59:33 2010 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 60E13106564A for ; Mon, 24 May 2010 05:59:33 +0000 (UTC) (envelope-from nvass9573@gmx.com) Received: from mailout-eu.gmx.com (mailout-eu.gmx.com [213.165.64.42]) by mx1.freebsd.org (Postfix) with SMTP id A6C208FC18 for ; Mon, 24 May 2010 05:59:32 +0000 (UTC) Received: (qmail invoked by alias); 24 May 2010 05:59:31 -0000 Received: from unknown (EHLO moby.local) [79.107.189.136] by mail.gmx.com (mp-eu003) with SMTP; 24 May 2010 07:59:31 +0200 X-Authenticated: #46156728 X-Provags-ID: V01U2FsdGVkX18Hs+c+vr4HJy+8wmBGCb5x0xRXOAL/4K7Y1d+e95 CHkM+8BaBkBDBQ Message-ID: <4BFA15BE.1060607@gmx.com> Date: Mon, 24 May 2010 08:59:26 +0300 From: Nikos Vassiliadis User-Agent: Thunderbird 2.0.0.23 (X11/20100313) MIME-Version: 1.0 To: Eitan Adler References: <4BE84825.9060005@gmx.com> <1a7012fe7affe8caf4263d4d2c385614.squirrel@whipp.no-ip.org> <4BF10F3D.2070207@gmx.com> In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Y-GMX-Trusted: 0 Cc: crwhipp@gmail.com, freebsd-questions@freebsd.org Subject: Re: File system X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 24 May 2010 05:59:33 -0000 Eitan Adler wrote: > gjournal will replay all write attempts >> (metadata and data) before the failure, so you should be relatively >> sure that all writes are done correctly. > > As I understand it journals work by writing to disk a log of all the > changes that have to be made - waits for confirmation that it wrote > the data - and then attempts to make those changes. If after the > confirmation there is a crash the log file is replayed. > Certain virtual machines will report to the OS that it wrote the data > to disk before it actually does so. In that case journaling doesn't > actually help as the log file is still not on some form of stable > storage. I am not an expert on the subject, I thought the journal will replay all logged write attempts and since the number of all write attempts logged in the journal will be much bigger than the number of requests a cache can hold you will be sure that all writes will be done on the filesystem. Again, I am not an expert on the subject... Nikos