Rules for Using the HA Database Monitor with ServiceGuard

Setting Failover Parameters

When using the HA Database Monitor with ServiceGuard, the ServiceGuard parameters listed in Table 3-3 “Setting Failover Parameters” should be set so that a package failover will occur when access to the database resource fails.

Sample File Settings

The following is an example of how you might set up an ServiceGuard cluster, mycluster, with two nodes, nestle and whitman, to monitor the availability of a database, db_1, that is defined on a volume group, VG01, that can be accessed in exclusive mode by either nestle or whitman. Figure 3-2 “Two Node Cluster with HA Database Monitor Failover” shows the sample cluster setup.

Figure 3-2 Two Node Cluster with HA Database Monitor Failover

This sample illustrates a single database failover situation. You can also configure failover for multiple databases, for example:

In the above scenario, ServiceGuard handles the activation and mounting of the Oracle database, as well as starting the database instance for you.

Setting Up the ServiceGuard Package

This section provides an example, based on Figure 3-2 “Two Node Cluster with HA Database Monitor Failover”, of setting the parameters in the ServiceGuard package configuration file. The sample lists the parameters for the HA Database Monitor, not all the parameters needed to set up ServiceGuard. For a complete listing and explanation of all the parameters, refer to your Managing MC/ServiceGuard (B3936-90024) book. Table 3-4 “Some ServiceGuard Package Configuration File Parameters” lists the ServiceGuard package configuration file parameters that are used to configure the sample two-node called mycluster that is depicted in Figure 3-2 “Two Node Cluster with HA Database Monitor Failover”.

Table 3-5 “Some ServiceGuard Package Control Script Parameters” lists the ServiceGuard package control script parameters for the shared volume group, VG01.

ServiceGuard Sample Run/Halt Scripts

The following is a sample of the run and halt portion of the ServiceGuard package control script.

In the section for customer-defined run commands, type:

function customer_defined_run_cmds
{
export ORACLE_HOME=<oracle_home_directory>
export ORACLE_SID=<oracle_sid>
PFILE=$ORACLE_HOME/dbs/init<sid>.ora
su oracle -c $ORACLE_HOME/bin/svrmgrl <<EOF
connect internal
startup pfile=${PFILE}
exit
EOF
test_run 51
}

In the section for customer-defined halt commands, type:

function customer_defined_run_cmds
{
export ORACLE_HOME=<oracle_home_directory>
export ORACLE_SID=<oracle_sid>
PFILE=$ORACLE_HOME/dbs/init<sid>.ora
su oracle -c $ORACLE_HOME/bin/svrmgrl <<EOF
connect internal
shutdown
exit
EOF
test_run 52
}

Points to consider:

By adding the Oracle commands into the run and halt portions of the configuration control script, you allow ServiceGuard to start up and shut down the Oracle instance for you.