From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755220AbYGVSgd (ORCPT ); Tue, 22 Jul 2008 14:36:33 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752722AbYGVSgZ (ORCPT ); Tue, 22 Jul 2008 14:36:25 -0400 Received: from qw-out-2122.google.com ([74.125.92.27]:13703 "EHLO qw-out-2122.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751223AbYGVSgY (ORCPT ); Tue, 22 Jul 2008 14:36:24 -0400 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:date:from:to:subject:mime-version:content-type :content-transfer-encoding:content-disposition; b=NPtJuCTiqmxA5omv5u5jwqSGMkXRDQwAaZvHnjT/8VtbcQYNYtKduUZ+P5eZXJl+ky ucQNDMhnFPQGzHQPrIME7tDOQDH3iRE1I5Ekkoj3KUsm84HwsrPBgjN1nOnmwsUUmlCj 8wPN+nK1fMOG3k3SIsC6DPWC3vZFBX9qYIWPA= Message-ID: <8fb5fa2d0807221136v2b92a34cv2c1fe4a5e9a126bc@mail.gmail.com> Date: Tue, 22 Jul 2008 11:36:22 -0700 From: "Buddy Lumpkin" To: linux-kernel@vger.kernel.org, linux-mm@kvack.org Subject: measuring memory pressure, expose freepages and zone watermarks under /proc perhaps? MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi All, Would it be possible to expose the value of the zone watermarks and freepages for each zone under /proc in order to present a clear picture of what the memory pressure on a system looks like? Running: watch -n1 cat /proc/meminfo, you can get a pretty good feel for how much memory pressure each zone is under by watching HighFree and LowFree fall to a certain point, and then increase in a cyclical fashion, but it would be absolutely wonderful if some metrics could be exposed that give us a better view into the amount of memory pressure a system is under. In my case, I would like to write an agent that distills this information in a way that allows users to set alarm thresholds for their applications. The allocstall metric in /proc/vmstat is nice for indicating whether a 2.6 kernel has fallen under extreme memory pressure, but I have applications that are affected noticeably by even modest amounts of memory pressure make a very noticeable difference in latency. Lastly, if anyone knows of a good method for measuring various levels of memory pressure on 2.4 and/or 2.6 kernels, please share them. Thanks in advance, --Buddy