From: Dave Chinner Subject: Re: filesystems bigger than 16 TB? Date: Mon, 27 Jun 2011 10:40:23 +1000 Message-ID: <20110627004023.GE32466@dastard> References: <4E03251F.4040603@wpkg.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org To: Tomasz Chmielewski Return-path: Received: from ipmail04.adl6.internode.on.net ([150.101.137.141]:16523 "EHLO ipmail04.adl6.internode.on.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756069Ab1F0Ak3 (ORCPT ); Sun, 26 Jun 2011 20:40:29 -0400 Content-Disposition: inline In-Reply-To: <4E03251F.4040603@wpkg.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Thu, Jun 23, 2011 at 01:35:59PM +0200, Tomasz Chmielewski wrote: > With mkfs.ext4 from 1.41.14, it is not possible to create a > filesystem which is bigger than 16 TB: > > > mkfs.ext4: Size of device /dev/sdb too big to be expressed in 32 bits > using a blocksize of 4096. > > > But I see it succeeds with the latest git version of e2fsprogs. > > > > The question is: how reliable such a filesystem is? Regardless of the filesystem or the feature, if it is not in officially released packages, do you really want to risk your production data on an experimental filesystem/feature? > On a system which is supposed to be reliable, perhaps I'll be better > off with xfs for such large filesystems? > > I'm using Debian Squeeze, which has a 2.6.32 kernel. On a 2.6.32 kernel, I'd strongly recommend using XFS for >16TB filesystems.... Cheers, Dave. -- Dave Chinner david@fromorbit.com