From: Jan Kara Subject: Re: old/new ext3 compatibility Date: Mon, 23 Mar 2009 19:34:28 +0100 Message-ID: <20090323183428.GE15488@atrey.karlin.mff.cuni.cz> References: <20090319184501.GG3634@mtholyoke.edu> <20090319220224.GM3634@mtholyoke.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org To: Ron Peterson Return-path: Received: from atrey.karlin.mff.cuni.cz ([195.113.26.193]:36294 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754369AbZCWSea (ORCPT ); Mon, 23 Mar 2009 14:34:30 -0400 Content-Disposition: inline In-Reply-To: <20090319220224.GM3634@mtholyoke.edu> Sender: linux-ext4-owner@vger.kernel.org List-ID: > 2009-03-19_14:45:02-0400 rpeterso: > > As I understand it, debian lenny's ext3 filesystem uses 256 byte inodes, > > to be forward compatible with ext4. > > > > I have a production server running debian etch. It is attached to a > > fiber channel array, on which it has several ext3 filesystems. I'm > > installing a new server, and I'd like to use lenny. It will be attached > > to the same array, and I'd like to be able to occasionally use the ext3 > > filesystems created previously. Ideally, I'd also like to go the other > > direction as well. Is this possible, or just crazy talk? > > If I understand what I'm reading correctly, this is a non-problem. Any > recent 2.6 kernel should understand ext3 filesystems with 256 byte > inodes just fine. The only thing that has happened is that the latest > debian stable defaults to using 256 byte inodes rather than 128. Is > that correct? Are there any gotcha's here that I should be aware of? Yes. Any 2.6 kernel will understand both inode sizes. It is just a matter of a default in mkfs. Larger inodes allow for more effective handling of extended attributes, ACLs and such but waste more space/throughput in case you don't use them... Honza -- Jan Kara SuSE CR Labs