From mboxrd@z Thu Jan 1 00:00:00 1970 From: Theodore Ts'o Subject: Re: [PATCH] generic/224: Increase filesystem instance size to 1.5 GiB Date: Mon, 31 Aug 2015 17:09:27 -0400 Message-ID: <20150831210927.GD27040@thunk.org> References: <1440945981-323-1-git-send-email-chandan@linux.vnet.ibm.com> <20150831181127.GB7642@thunk.org> <55E4A8BA.2050908@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Chandan Rajendra , fstests@vger.kernel.org, linux-btrfs@vger.kernel.org, linux-fsdevel@vger.kernel.org, chandan@mykolab.com To: Austin S Hemmelgarn Return-path: Content-Disposition: inline In-Reply-To: <55E4A8BA.2050908@gmail.com> Sender: fstests-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Mon, Aug 31, 2015 at 03:19:22PM -0400, Austin S Hemmelgarn wrote: > AFAIK, it shouldn't be failing that way, and should automatically switch to > mixed mode allocation. A 1G filesystem should work fine for BTRFS, but > smaller ones will have higher chances of ENOSPC issues (inversely > proportional to the size of the FS). I would advise against using BTRFS on > such a small disk (I avoid using it on anything smaller than 4G personally), > but I'm not one of the developers, and the fact that I feel it isn't a good > idea doesn't mean it shouldn't work. Instead of modifying generic/224, maybe it would be better to have a way to specify a minimum file system size on a per-file system basis. That way, if some file system does have a minimum size of say, 1G or 4G, it can be configured in one place, instead of needing to modify every test that uses a small file system size, or forcing all file systems to use a larger file systems just for the benefit of a single file system? Or maybe just fix mkfs.btrfs? - Ted