From mboxrd@z Thu Jan 1 00:00:00 1970 From: Suparna Bhattacharya Subject: Re: [RFC][PATCH] ChunkFS: fs fission for faster fsck Date: Mon, 23 Apr 2007 21:58:49 +0530 Message-ID: <20070423162849.GB14113@in.ibm.com> References: Reply-To: suparna@in.ibm.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, val_henson@linux.intel.com, riel@surriel.com, zab@zabbo.net, arjan@infradead.org, brandon@ifup.org, karunasagark@gmail.com, gud@ksu.edu To: Amit Gud Return-path: Received: from e6.ny.us.ibm.com ([32.97.182.146]:47137 "EHLO e6.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754038AbXDWQVt (ORCPT ); Mon, 23 Apr 2007 12:21:49 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Mon, Apr 23, 2007 at 06:21:34AM -0500, Amit Gud wrote: > > This is an initial implementation of ChunkFS technique, briefly discussed > at: http://lwn.net/Articles/190222 and > http://cis.ksu.edu/~gud/docs/chunkfs-hotdep-val-arjan-gud-zach.pdf > > This implementation is done within ext2 driver. Every chunk is an > independent ext2 file system. The knowledge about chunks is kept within > ext2 and 'continuation inodes', which are used to allow files and > directories span across multiple chunks, are managed within ext2. > > At mount time, super blocks for all the chunks are created and linked with > the global super_blocks list maintained by VFS. This allows independent > behavior or individual chunks and also helps writebacks to happen > seamlessly. > > Apart from this, chunkfs code in ext2 effectively only provides knowledge > of: > > - what inode's which block number to look for, for a given file's logical > block number > - in which chunk to allocate next inode / block > - number of inodes to scan when a directory is being read > > To maintain the ext2's inode number uniqueness property, 8 msb bits of > inode number are used to indicate the chunk number in which it resides. > > As said, this is a preliminary implementation and lots of changes are > expected before this code is even sanely usable. Some known issues and > obvious optimizations are listed in the TODO file in the chunkfs patch. > > http://cis.ksu.edu/~gud/patches/chunkfs-v0.0.8.patch > - one big patch > - applies to 2.6.18 Could you send this out as a patch to ext2 codebase, so we can just look at the changes for chunkfs ? That might also make it small enough to inline your patch in email for review. What kind of results are you planning to gather to evaluate/optimize this ? Regards Suparna > > Attached - ext2-chunkfs-diff.patch.gz > - since the code is a spin-off of ext2, this patch explains better what > has changed from the ext2. > > git://cislinux.cis.ksu.edu/chunkfs-tools > - mkfs, and fsck for chunkfs. > > http://cis.ksu.edu/~gud/patches/config-chunkfs-2.6.18-uml > - config file used; tested mostly on UML with loopback file systems. > > NOTE: No xattrs and xips yet, CONFIG_EXT2_FS_XATTR and CONFIG_EXT2_FS_XIP > should be "no" for clean compile. > > > Please comment, suggest, criticize. Patches most welcome. > > > Best, > AG > -- > May the source be with you. > http://www.cis.ksu.edu/~gud -- Suparna Bhattacharya (suparna@in.ibm.com) Linux Technology Center IBM Software Lab, India