From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Wed, 1 May 2002 06:02:44 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Wed, 1 May 2002 06:02:43 -0400 Received: from smtp1.libero.it ([193.70.192.51]:37512 "EHLO smtp1.libero.it") by vger.kernel.org with ESMTP id convert rfc822-to-8bit; Wed, 1 May 2002 06:02:42 -0400 Content-Type: text/plain; charset="us-ascii" From: Andrea Aime To: linux-kernel@vger.kernel.org Subject: How to tune vm to be less swap happy? Date: Wed, 1 May 2002 12:02:38 +0000 User-Agent: KMail/1.4.5 MIME-Version: 1.0 Content-Transfer-Encoding: 8BIT Message-Id: <200205011202.38653.aaime@libero.it> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi everybody, sorry to bother you with this question, but I feel that current vm defaults are not tuned for a workstation use, but for a file server/db server use. I work on a couple of machines with 128MB of RAM, tipically use KDE 3.0 and Netbeans (a Java IDE). Ok, I use bloated software, I should have more RAM. Fine. But what I don't understand is why the system keeps swapping data in and out just because it doesn't want to shrink the cache below 40 MB. My typical memory occupation is: total used free shared buffers cached Mem: 127096 123264 3832 0 1216 48164 -/+ buffers/cache: 73884 53212 Swap: 257000 62316 194684 Now, as you can see I would need 135 MB to keep all my programs in memory, and if you look at top output: PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND 4397 wolf 9 0 79196 71M 30472 S 0,0 57,4 0:00 java 4398 wolf 8 0 79196 71M 30472 S 0,0 57,4 0:00 java ... you'll notice that due to the cache behaviour not even netbeans fits enterely in memory. I really doubt that I can benefit from a 48MB disk cache since the system is swapping in and out every time I open konqueror, kmail and the like... is there anything I can do to shrink the cache and to allow a bigger part of my working set to fit into memory? Moreover, what vm tuning do you advice for a typical workstation use with memory hungry applications? Best regards Andrea Aime