Implementation of DB2 dual-machine ha complete scheme on Linux

1. The SUMMARY

This document describes the basic steps for installing a high reliability scheme for configuring DB2 UDB Enterprise Serverv8.2 dual-machine SuSE Linux Enterprise Server v8.0 (SLES 8). The scheme is implemented as an HA implementation component in conjunction with the SLEs Volume Manager (LVM) and Veritas Cluster Server v2.2 (VCS 2.2).

2. Overview

This document assumes that the reader has understood the basic concepts of the HA scheme for dual-machine preparation.

2.1. The basic steps of two-machine mutual-standby ha scheme

The basic steps to establish a dual-machine interoperability plan are:

1. Determine the basic parameters (such as IP address, storage space, etc.). The parameters of this scheme are all model parameters, and the readers need to replace them according to the actual environment.

2. Configure shared storage (this scenario does not involve the configuration of shared storage scenarios)

3. Install the application separately on both nodes (DB2 UDB in this scenario)

4. Initial configuration of the application on a single node (in this scenario, a DB2 database is established on shared storage)

5. Introduce the configuration on the shared storage on another node (in this scenario, do a catalog operation on the database on shared storage)

6. Manually test the application on two nodes

7. Configure HA cluster management software (VCS in this scenario)

8. Test HA cluster management software to successfully take over resources

2.2. Suppose

This document assumes the following example environment in which SLEs and VCs are properly installed on the node, SLEs shared storage is properly connected, and the VCs ' heartbeat connection is properly configured.

2.2.1. Storage

Sufficient local disk space is required on each node to install executable code and instances for DB2 UDB.

This scheme requires enough shared storage space to place the database data.

Assume that the volume group named for the DB2 database on the shared storage is/DEV/DATAVG1, the logical volume name is/DEV/DATAVG1/DB2LV1 (the LVM service using SLEs), and the hanging point on two nodes is named/home/db2data, and has been properly formatted as the appropriate file system (Veritas engineers point out that the VCS 2.2 does not support the Ext3 file system on SLEs 8, it is recommended that you specify ext2 when formatting, and this problem does not exist on SLEs 9 and VCs 4.1). Note that this hook is configured to not automatically hook up in the Fstab file when it is started.

Other table spaces in the database can be built on other volumes of shared storage, and if it is a file system, it is also configured to not automatically hook up. This document is not discussed at this stage.

2.2.2. Node

This HA scheme is made up of two server nodes, which are called active nodes and passive nodes respectively. These two nodes have the same hardware and operating system configuration.