From: Theodore Ts'o Subject: Re: [RFC PATCH v2 0/4] ext4: extents status tree shrinker improvement Date: Wed, 16 Apr 2014 11:19:38 -0400 Message-ID: <20140416151938.GA17208@thunk.org> References: <1397647830-24444-1-git-send-email-wenqing.lz@taobao.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org, Zheng Liu , Andreas Dilger , Jan Kara To: Zheng Liu Return-path: Received: from imap.thunk.org ([74.207.234.97]:55162 "EHLO imap.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1161737AbaDPPTo (ORCPT ); Wed, 16 Apr 2014 11:19:44 -0400 Content-Disposition: inline In-Reply-To: <1397647830-24444-1-git-send-email-wenqing.lz@taobao.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: Hi Zheng, Thanks so much for your deep and detailed work on this. I have two immediate reactions. 1) We should first fix __ext4_es_shrink so that we interpret nr_to_scan correctly --- it's the number of objects to scan, not the number of objects that we need to shirnk. That should significantly reduce the number of scans that we do, and fixing this could potentially influence the metrics that we measure. 2) In addition to measuring the scan time, we should also measure how many times we end up loading extents from disk when (during the course of a normal workload) we are under memory pressure, and RR ends up throwing out an extent that would have been saved in LRU, and so an additional metadata read would be required when using RR. - Ted