Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264448AbUDZKGR (ORCPT ); Mon, 26 Apr 2004 06:06:17 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264479AbUDZKGM (ORCPT ); Mon, 26 Apr 2004 06:06:12 -0400 Received: from kilobug.org ([62.212.118.231]:642 "EHLO drizzt.home") by vger.kernel.org with ESMTP id S264471AbUDZKGE (ORCPT ); Mon, 26 Apr 2004 06:06:04 -0400 To: linux-kernel@vger.kernel.org Subject: SMP, Soft RAID and abnormal load Mail-Copies-To: nobody From: kilobug@freesurf.fr (=?iso-8859-1?q?Ga=EBl_Le_Mignot?=) Organization: HurdFr - http://hurdfr.org X-PGP-Fingerprint: 1F2C 9804 7505 79DF 95E6 7323 B66B F67B 7103 C5DA Date: Mon, 26 Apr 2004 12:06:03 +0200 Message-ID: User-Agent: Gnus/5.1006 (Gnus v5.10.6) Emacs/21.3 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2204 Lines: 57 Hello, We are running a MMO game server (made of several multi-threaded porcessus using CORBA to communicate) on a computer with two Xeon processor and a software RAID level 1 between two ide discs (hda and hdc). The load average is usually around 0.5: 18:58:11 up 42 days, 1 min, 6 users, load average: 0.30, 0.41, 0.51 We tried to unzip a zip file containing a lot of small files, and then the load average increased dramatically, up to more than 23. I'm wondering how a single processus can increase the load average that much. The kernel is vanilla 2.4.25, with SMP, highmem, and sw raid: Linux mogii1 2.4.25-raid #1 SMP Tue Feb 24 14:37:33 CET 2004 i686 unknown The filesystem is ext3fs, mounted with the default "ordered" mode. The system has plenty of memory available: total used free shared buffers cached Mem: 3879840 3750980 128860 0 299948 1709776 -/+ buffers/cache: 1741256 2138584 Swap: 1959848 83964 1875884 I didn't fine-tune the discs with hdparm (this is a production server, and I don't want to take any risk of data corruption), here is the report of hdparm: multcount = 16 (on) I/O support = 0 (default 16-bit) unmaskirq = 0 (off) using_dma = 1 (on) keepsettings = 0 (off) nowerr = 0 (off) readonly = 0 (off) readahead = 8 (on) geometry = 14946/255/63, sectors = 240121728, start = 0 busstate = 1 (on) If you have an explication on why did it happen, I'ld be gratefull. I hope this report may help you to "tune" the kernel. If you need additional informations, don't hesitate to ask. -- Gael Le Mignot "Kilobug" - kilobug@nerim.net - http://kilobug.free.fr GSM : 06.71.47.18.22 (in France) ICQ UIN : 7299959 Fingerprint : 1F2C 9804 7505 79DF 95E6 7323 B66B F67B 7103 C5DA Member of HurdFr: http://hurdfr.org - The GNU Hurd: http://hurd.gnu.org - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/