From: Theodore Tso Subject: Re: Something wrong with extent-based journal creation Date: Wed, 17 Jun 2009 11:31:26 -0400 Message-ID: <20090617153126.GA25451@mit.edu> References: <4A381EC8.4010202@redhat.com> <20090617080432.GC13073@webber.adilger.int> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Eric Sandeen , ext4 development To: Andreas Dilger Return-path: Received: from THUNK.ORG ([69.25.196.29]:52349 "EHLO thunker.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752830AbZFQPbo (ORCPT ); Wed, 17 Jun 2009 11:31:44 -0400 Content-Disposition: inline In-Reply-To: <20090617080432.GC13073@webber.adilger.int> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Wed, Jun 17, 2009 at 02:04:32AM -0600, Andreas Dilger wrote: > On Jun 16, 2009 17:38 -0500, Eric Sandeen wrote: > > I've narrowed it down to the commit (961306d3) which creates the journal > > with extent format; if that's commented out, it works fine. > > Hmm, this seems like it might be a troublesome feature... Are we sure > that there is proper kernel/tool compatibility for an extent-mapped > journal? At least none of the Lustre code has any support for this, > though they can handle the INCOMPAT_EXTENTS feature, and I don't think > this was present in older kernels either. > > I haven't looked at the code for this yet, so it may be that it will > "just work", but if this is a new feature I would urge caution w.r.t. > compatibility before this is merged upstream (in either e2fsprogs or > the kernel). Extent-based journals "just works", at least in the kernel; I didn't need to make any changes to the kernel before I added support to using an extent-mapped journal; if the kernel supports extents, it should expert extent-mapped journals without any difficults. The patch to add this was merged in before e2fsprogs 1.41.1 was released, so mke2fs has been creating extent-based journals for a while. - Ted