# "@(#)41 1.7 src/43haes/usr/sbin/cluster/sa/oracle/sbin/README, hacmp.assist.oracle, 61haes_r714 6/29/05 09:25:38" # IBM_PROLOG_BEGIN_TAG # This is an automatically generated prolog. # # 61haes_r714 src/43haes/usr/sbin/cluster/sa/oracle/sbin/README 1.7 # # Licensed Materials - Property of IBM # # COPYRIGHT International Business Machines Corp. 2005 # All Rights Reserved # # US Government Users Restricted Rights - Use, duplication or # disclosure restricted by GSA ADP Schedule Contract with IBM Corp. # # IBM_PROLOG_END_TAG This is the README for the Oracle Assist program. The assist configures an HACMP two node cluster and helps install and configure the Oracle Infrastructure Tier software. It is assumed that this program is being run by root from the primary node. You will need to provide the hostname of the second node, and the HACMP service IP address. See the Install Oracle section below for requirements specific to Oracle. The Oracle emctl requires that the DISPLAY environment variable be non-null on the cluster nodes. There is one optional commandline flag, -v. This Verbose flag will chart the progress and display some results. The options file contains default values for some attributes of the assist. The user may customize this file to increase Filesystem sizes, change the application start and stop script locations, or alter other attributes. There is a log file, and a backup log file, /var/hacmp/log/oracle_sa.log and /var/hacmp/log/oracle_sa.log.bak Other useful logs are /tmp/hacmp.out /tmp/cspoc.out /var/hacmp/clverify/clverify.log /smit.log These are the tasks to create the two node cluster. system check - This task examines the Swap(paging) Space, /usr, /tmp and /var free space that is required for the Oracle installation. configure basic cluster - If there is no cluster defined, this task will create a two node cluster using the node you are running on as the primary node. You will need to enter the hostname for the secondary node. You can perform this task Automatically or by using SMIT. It will discover the available networks and volume groups. It will create the user "oracle" and the group "dba" on the cluster nodes. prompt and validate vg - You can choose an existing shared volume group or select "new_name" and create a new shared volume group. It then creates a volume group or validates an existing volume group. If you are creating a shared volume group, you will need to select the shared physical volumes that will be in the volume group. Select from the list, one at a time. Hit Enter without selecting when done. This volume group will be recognized on the secondary node. You can choose whether to use one or two disks. Oracle recommends two disks, one for the database, one for the commands. If you are using luns (san array) you might prefer one disk. prompt and validate fs - Select the filesystem names you wish to use on the shared volume group. Oracle requires 4.3GB of space, which is 860,000 512 blocks. The assist will create 2 Logical Volumes and two Filesystems, with 5 GB each. They will be located on the disk(s) that you selected for the volume group. After the filesystems are verified, they are mounted. configure application server - Select a name for HACMP to reference the Infrastructure Tier application and the scripts that start and stop it. If you choose automatic for this task, it will use the scripts that are shipped with the assist. You will be put into "vi" so you can insure that they match the software levels and the ORACLE_HOME that you are using for the Oracle install process. define service IP label - The service IP address can be configured in a resource group to make it highly available across multiple nodes. This is referred to as the Virtual IP (VIP) in Oracle documentation. In addition to the IP address, you will need to specify the network on which this IP address will be bound. configure resource group - This task will configure the Resource Group for this cluster. If there is no resource group for this cluster, you will create one. Configure the resource group by adding the volume group, service IP_label and the application server. define private network - Oracle requires a private network, select one from the list of networks that have been discovered. You can perform this automatically or by choosing SMIT. If you use SMIT, be sure to select the Network attribute "private". This network cannot be the network on which the service IP label/Address will be bound. verify and sync - This task will verify the cluster configuration and attempt to "auto-correct" any problems that are found. It then syncs the configuration with the secondary node. You may choose to perform this automatically or with SMIT. If you use SMIT, select "Interactive" to correct errors found during verification. start hacmp - This task will start HACMP cluster services which makes the cluster ready for Oracle installation. You may do this automatically, or with SMIT. If you select SMIT, select both nodes for "Start Cluster Services on these nodes" install Oracle - Once the HACMP cluster is configured and running, Oracle can be installed. It requires a user group called "dba", and a user called "oracle". If they do not exist, create them with smitty cl_usergroup. Oracle uses the 1521 port, if it is not free, you will be notified. Then enter the DISPLAY variable so that the "Oracle Universal Installer" can run on your terminal. Oracle is shipped with a program called rootpre.sh, which has to run on both nodes. The program needs to know the location of this script. You may find it on the Oracle Disk1. This is also the location of "runInstaller". Oracle also requires the "Java Cryptography Extension" in jce-1_2_2.zip, which is available from the Java web site. rootpre.sh checks the system for space and software requirements specific to Oracle. runInstaller will start the "Oracle Universal Installer" (OUI) GUI. Be sure to select "CFC Cold Failover Cluster" during the installation, and select shared filesystems for the commands directory, inventory directory, and the database directory. You can run the OUI installer without the assist by using the "runInstaller -ignoreSysPrereqs" command as user oracle. Starting the Oracle Infrastructure Tier - After the Oracle installation process completes, it is started by stopping and starting the cluster from the smitty cl_cm_startstop_menu panel.