From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263868AbUDONrd (ORCPT ); Thu, 15 Apr 2004 09:47:33 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263983AbUDONrd (ORCPT ); Thu, 15 Apr 2004 09:47:33 -0400 Received: from webmail2.speakeasy.net ([216.254.0.82]:6026 "EHLO webmail.speakeasy.net") by vger.kernel.org with ESMTP id S263868AbUDONr3 (ORCPT ); Thu, 15 Apr 2004 09:47:29 -0400 From: ward@speakeasy.net To: fedora-test-list@redhat.com Cc: linux-kernel@vger.kernel.org Importance: Normal Sensitivity: Normal Message-ID: X-Mailer: Mintersoft VisualMail, Build 4.0.111601 X-Originating-IP: [199.196.144.16] Date: Thu, 15 Apr 2004 13:47:27 +0000 Subject: FC2T2 shared memory segments cannot be allocated by DB2 V8.1 MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org I'm running into regular memory related issues with a Fedora Core 2 system which appears to have been properly configured using generous sysctl.conf settings. The symptoms are that I can create a limited number of connections to unique databases running on this development system and then after about the fifth or sixth I encounter an error indicating a failure to allocate shared memory segments. Running ipcs -m at the time of the error I can see 8 shared memory segments owned by that userid and 32 total on the system. ipcs -l indicates the settings I've made to sysctl.conf... ------ Shared Memory Limits -------- max number of segments = 4096 max seg size (kbytes) = 1048576 max total shared memory (kbytes) = 8388608 min seg size (bytes) = 1 The database engine indicates this error in its logs as an occurance of a Memory allocation failure. Unsure if this was related to a similar issue which impacts Oracle I explored various combinations of kernels and patches including support for the patch to add a sysctl capability to control the disable_cap_mlock feature. Note that after the occurance of a failure I'm experiencing problems with the forking of processes. This was seen in kernel compiles and while halting the system. Any ideas?