From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751386AbWASJgf (ORCPT ); Thu, 19 Jan 2006 04:36:35 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751389AbWASJgf (ORCPT ); Thu, 19 Jan 2006 04:36:35 -0500 Received: from mail09.syd.optusnet.com.au ([211.29.132.190]:55181 "EHLO mail09.syd.optusnet.com.au") by vger.kernel.org with ESMTP id S1751386AbWASJge (ORCPT ); Thu, 19 Jan 2006 04:36:34 -0500 From: Con Kolivas To: Andrew Morton Subject: Re: Out of Memory: Killed process 16498 (java). Date: Thu, 19 Jan 2006 20:36:20 +1100 User-Agent: KMail/1.9 Cc: "Andy Chittenden" , linux-kernel@vger.kernel.org References: <89E85E0168AD994693B574C80EDB9C2703555F85@uk-email.terastack.bluearc.com> <20060119005600.4e465e9d.akpm@osdl.org> In-Reply-To: <20060119005600.4e465e9d.akpm@osdl.org> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200601192036.20355.kernel@kolivas.org> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thursday 19 January 2006 19:56, Andrew Morton wrote: > "Andy Chittenden" wrote: > > Why does running the following command cause processes to be killed: > > > > dd if=/dev/zero of=/u/u1/andyc/tmpfile bs=1M count=8k > > > > And I noticed one of my windows disappeared. Further investigation > > showed that was my terminator window (java based app: see > > http://software.jessies.org/terminator/). I found this in my syslog: > > > > Jan 17 11:12:58 boco kernel: Out of Memory: Killed process 16498 (java). > > > > My hardware: amd64 based machine (ASUS A8V Deluxe motherboard) with 4Gb > > of memory. > > My kernel: debian package linux-image-2.6.15-1-amd64-k8 package > > installed. IE its running 2.6.15 compiled for amd64. > > > > This is repeatable. The above dd command also causes the machine to > > become very unresponsive (eg windows don't focus). > > What type of filesytem is being written to? > > Has someone tuned /proc/sys/vm/swappiness, or something else under > /proc/sys/vm? > > It'd be useful to see the dmesg output from that oom event. Are you using scsi? Someone just posted what looks to be a scsi slab leak (Re: scsi cmd slab leak? (Was Re: [ck] Anyone been having OOM killer problems lately?) that causes oom kills. Check your slabinfo for a large scsi_cmd_cache. Cheers, Con