From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Fri, 08 Dec 2000 09:09:57 -0500 From: Chris Mason Subject: Re: [linux-lvm] How do you mount a snapshot? Message-ID: <323270000.976284597@coffee> In-Reply-To: <20001207223300.D25533@archimedes.oak.suse.com> MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline Sender: linux-lvm-admin@sistina.com Errors-To: linux-lvm-admin@sistina.com Reply-To: linux-lvm@sistina.com List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: List-Id: Content-Type: text/plain; charset="us-ascii" To: linux-lvm@sistina.com On Thursday, December 07, 2000 22:33:00 -0800 David Gould wrote: > On Thu, Dec 07, 2000 at 03:56:11PM -0500, Chris Mason wrote: [ snapshots ] >> It should work with ext2. > > Really? I wonder. Maybe right after a sync before anything gets dirtied, > but otherwise, how could the disk image be consistantly, er, consistant? > And would not e2fsck have the same issues about writing? > > Or did I just wake up pessimistic today? > The reason it will not work on reiserfs without the extra API is that on reiserfs, sync == commit to log, not commit to disk. So, there is no way over the command line (or in the kernel) to ask the FS to sync to disk, even when there is no traffic at all on the FS. With ext2, if nothing is going on, you can type sync and have all the buffers on disk. It is at least possible ;-) It would be better if ext2 could provide the write suspension hooks as well, but I'm guessing that will be done in ext3 first (where the journal already does most of the work). In my mind, the people that need snapshots the most are the database programs, where disk traffic can be carefully controlled. So the current ext2 method is still useful. -chris