From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Darrick J. Wong" Subject: Re: resize2fs: Should never happen: resize inode corrupt! - lost key inodes Date: Fri, 18 Sep 2015 22:23:46 -0700 Message-ID: <20150919052346.GF10390@birch.djwong.org> References: <20150811224718.GD20658@thunk.org> <55CBC212.6000507@harvyl.se> <20150813132716.GB26095@thunk.org> <55CCDE20.3010108@harvyl.se> <55E8C6AC.5000308@harvyl.se> <55F3FE07.9030807@harvyl.se> <55F73DAD.6020807@harvyl.se> <55F85B90.3000709@harvyl.se> <20150919024725.GV3902@dastard> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Andreas Dilger , Johan Harvyl , "Theodore Ts'o" , "linux-ext4@vger.kernel.org" To: Dave Chinner Return-path: Received: from userp1040.oracle.com ([156.151.31.81]:26705 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751343AbbISFXy (ORCPT ); Sat, 19 Sep 2015 01:23:54 -0400 Content-Disposition: inline In-Reply-To: <20150919024725.GV3902@dastard> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Sat, Sep 19, 2015 at 12:47:25PM +1000, Dave Chinner wrote: > On Wed, Sep 16, 2015 at 07:21:59PM -0600, Andreas Dilger wrote: > > If you add "-b 1024" to the mke2fs command line to use 1KB instead of 4KB blocks, and reduce the sizes by a factor of 4 does the problem still happen? That would make it easier for someone else to test, since it would only need a 4-5TB disk instead of a 19Tb array. > > Sparse files on XFS using loopback will allow you to simulate > devices larger than 16TB easily. You can turtle it all the way down, > too, to create the xfs filesystem on a loopback device on a sparse > file on ext4.... > > Doing this sort of thing lets me know, for example, that the > mkfs.ext4 defaults fail on a 500TB device... > > # xfs_io -f -c 'truncate 500t' /mnt/xfs/fs.img > # ls -lh /mnt/xfs > total 0 > -rw------- 1 root root 500T Sep 19 12:41 fs.img > # mkfs.ext4 /mnt/xfs/fs.img > mke2fs 1.42.13 (17-May-2015) > /mnt/xfs/fs.img: Cannot create filesystem with requested number of inodes while setting up superblock Whee. I guess one would need to turn on meta_bg at mkfs time (which scatters the group descriptors across the disk instead of (failing to) sandwich them in a single blockgroup... and fix the overhead calculation in ext2fs_initialize to calculate the maximum BG overhead correctly, since it doesn't seem to know about metabg. Of course there's the question of whether or not we really /want/ people formatting 500T ext4 filesystems. meta_bg is not turned on by default, so the defaults will still fail unless they know to pass that option. (Frankly, doing so is probably insane.) --D > # > > Cheers, > > Dave. > -- > Dave Chinner > david@fromorbit.com > -- > To unsubscribe from this list: send the line "unsubscribe linux-ext4" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html