From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1765772AbXGVLMS (ORCPT ); Sun, 22 Jul 2007 07:12:18 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754728AbXGVLMG (ORCPT ); Sun, 22 Jul 2007 07:12:06 -0400 Received: from [212.12.190.99] ([212.12.190.99]:33634 "EHLO raad.intranet" rhost-flags-FAIL-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1753576AbXGVLMD (ORCPT ); Sun, 22 Jul 2007 07:12:03 -0400 From: Al Boldi To: linux-kernel@vger.kernel.org Subject: Re: [PATCH 0/3] readahead drop behind and size adjustment Date: Sun, 22 Jul 2007 14:11:59 +0300 User-Agent: KMail/1.5 MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Message-Id: <200707221411.59594.a1426z@gawab.com> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Fengguang Wu wrote: > On Sun, Jul 22, 2007 at 10:24:37AM +0200, Peter Zijlstra wrote: > > On Sun, 2007-07-22 at 16:10 +0800, Fengguang Wu wrote: > > > - It will avoid large-file-reads-thrashing-my-desktop problem, > > > so most desktop users should like it. But sure there will be counter > > > cases when a user want to keep the data cached. > > > - File servers may hurt from it. Imagine a mp3/png file server. The > > > files are large enough to trigger drop-behind, but small (and hot) > > > enough to be cached. Also when a new fedora DVD iso is released, it > > > may be cached for some days. These are only the obvious cases. > > > > > > So I opt for it being made tunable, safe, and turned off by default. > > > > I'm still not convinced (Rik wasn't either last time around). When these > > files really are hot, they will be kept in memory due to them becoming > > Active. > > > > Also, by scaling up the max readahead size it takes a larger file before > > it starts dropping. If say this server has 4G of memory (not much at all > > for a server) resulting in a 1M readahead window, the file needs to be > > > ~2M before it starts drop behind. > > [snip] > > > But I guess it takes someone to try this IRL before we can settle this > > debate :-/ > > Yeah, some real workload numbers would help. A patch against 2.6.22 may help too. Thanks! -- Al