Configuring failover monitoring

Use the following task to set up failover monitoring. When the failover instance comes online during a failover, you can configure ORACLE_FAILOVER to automatically disable the instance on the secondary machine to avoid InstanceStatus going into an alarm state.

Before you begin

You must meet the following requirements:

• A PATROL Agent, Oracle, and PATROL for Oracle are running on the primary instance. You should use the same user name and password for PATROL on the failover instance and the primary instance.
• You are connected to and monitoring the primary instance from PATROL.
• A PATROL Agent and PATROL for Oracle are installed on the failover instance.
• The PATROL Agent is running on the failover instance when you set up failover monitoring on the primary instance.

To configure a failover instance (without loading the ORACLE_FAILOVER.kml)

1. From the ORACLE_INSTANCE application menu for the primary instance, choose KM Admin> Failover Monitoring> Configure.
2. In the dialog box, enter the following information:
   • The host name of the failover instance
   • The PATROL Agent port number for the failover instance
   • The Oracle SID of the failover instance
   • The Oracle home directory of the failover instance

3. Click Accept.
   The Configure [instance] for Failover Monitoring dialog box is displayed.

   Note

   This dialog box appears only the first time that you configure a failover instance.

4. Provide the following information and click Accept:

   
   
   Configure [ instance ] for Failover Monitoring dialog box fIelds

   Field	Description
   Backup Host Name	enter the computer name for the failover instance
   Backup Agent Port	enter the port number for the PATROL Agent on the failover instance computer
   Backup Oracle SID	enter the Oracle SID for the failover instance
   Backup Oracle Home	enter the Oracle home directory for the failover instance

   
   A link is established, and all PATROL for Oracle configuration information is transferred to the failover computer.

5. Verify that the following pconfig variables have been successfully transferred to the failover computer:
   • All configuration settings, including category and 24x7 monitoring
   • All exclusions for database objects, users, and tablespaces

   • all users selected for monitoring

     Note

     Repeat above steps from the procedure on each instance you want to configure for failover.

Configuring advanced failover monitoring

When the primary instance goes down, and the failover instance comes online, the InstanceStatus parameter for the failover instance goes into the OK state. The InstanceStatus parameter for the primary instance goes into an alarm or warning state, depending on the instance's online 24x7 status.

If you do not want the InstanceStatus parameter to go into an alarm state when a failover occurs, you can configure automatic monitoring on the failover computer by using the ORACLE_FAILOVER application, which is advanced failover monitoring.

You can implement advanced failover monitoring with one of the following parameters:

• FOBGPMonitor parameter (the default option requiring no setup). See Advanced failover monitoring with the FOBGPMonitor parameter (with the ORACLE_FAILOVER.kml).

• FOMonitor parameter (an option requiring setup). See Advanced failover monitoring with the FOMonitor parameter.

Advanced failover monitoring with the FOBGPMonitor parameter (with the ORACLE_FAILOVER.kml)

To configure advanced failover monitoring with the FOBGPMonitor parameter

1. Load ORACLE_FAILOVER.kml on the primary and failover computers.
2. Verify that in pre-discovery, the following conditions are true:
   • PATROL for Oracle is installed.
   • There are instances configured for failover by using the Instance > KM Admin > Failover Monitoring > Configure menu command.
     When these conditions are true, the AuxFailoverMonitor application instance appears under the ORACLE application.
3. From the Oracle icon menu of the failover host, choose KM Commands > KM Admin > Configuration > Advanced > Monitor Enable/Disable command, select the instance to exclude, and then click Apply.
   After discovery is complete, the standard parameter, FOBGPMonitor, runs every sixty seconds. The parameter looks for Oracle instances that are running:
   1. by searching for the dbw* background processes, and then
   2. by parsing instance names from those process names.
4. Using this process list, the parameter determines whether the instance under consideration is running on the current host:
   • If the instance is not running on the current host, and the instance was running during the previous poll cycle, the following actions occur:
   1. Monitoring for that instance is disabled.
   2. The parameter goes into a warning state until its next poll cycle.

   3. If the instance is still not running during the next poll cycle, the parameter is reset to the OK state.

      Note

      If you want to be notified of such an occurrence, you can set up notifications.

   • If the instance is running, but was not running during the previous poll cycle, the following actions occur:
   1. Monitoring for that instance is enabled

   2. The parameter goes into a warning state for one poll cycle before being reset.

      Note

      If you want to be notified of such an occurrence, you can set up notifications.

      
      
      These events occur in pairs: one indicating the instance going down on one host and another indicating the instance coming up on the other host. One or the other showing up alone indicates a problem.

Advanced failover monitoring with the FOMonitor parameter

Failover environments such as HP ServiceGuard and Solaris Clustering allow for the definition of packages or logical hosts. In this case, an application and the disks it uses are associated with a package. When a failover occurs, the relevant disks and packages are unmounted from the first host and mounted to the host to which the application moves.

The FOMonitor parameter looks for one of the application disks and uses the visibility of that disk to determine whether to activate monitoring.

This is a slightly stronger monitoring option than the monitoring option provided by the FOBGPMonitor parameter. The FOMonitor parameter determines where the instance should be running based on the presence of the disk, whereas the FOBGPMonitor parameter makes its determination based on where the instance is seen. If the instance is down on all hosts, no monitoring occurs at all. The InstanceStatus parameter could, for example, go into alarm if the disk is mounted and the instance did not come up.

However, the FOMonitor parameter requires setting up. You need to indicate the directory for the parameter to use for each instance. This can be the Oracle home. However, if the Oracle installation is on a common disk or is duplicated across local disks and is not part of the disks defined in the package, defining the Oracle home as the directory does not work. For information about setting up the FOMonitor parameter, see To configure advanced failover monitoring with the FOMonitor parameter.

To configure advanced failover monitoring with the FOMonitor parameter

1. Load ORACLE_FAILOVER.kml on the primary and failover computers.
2. Verify that in pre-discovery, the following conditions are true:
   • PATROL for Oracle is installed.
   • There are instances configured for failover by using the Instance > KM Admin > Failover Monitoring > Configure menu command.
     When these conditions are true, the AuxFailoverMonitor application instance appears under the ORACLE application.
3. From the Oracle icon menu of the failover host, choose KM Commands > KM Admin > Configuration > Advanced > Monitor Enable/Disable command, select the instance to exclude, and then click Apply.
4. Deactivate the FOBGPMonitor parameter under the AuxFailoverMonitor icon on both the primary and the failover computers.
5. Activate the FOMonitor parameter on both the primary and the failover computers.
6. From the host name of the primary instance, right-click and choose Development > Agent Configuration.
   The pconfig window is displayed.
7. In the pconfig window, choose Edit > Add Variable.
   The Add Variable dialog box is displayed.
8. In the Variable field, enter /OracleConfig/< ora_sid >/FailoverMonitoring/siddirpairslist.
   Where < ora_sid > is the Oracle instance identifier.
9. In the Value field, enter failover instance names and their respective directories in the format SID:directory_name, where SID denotes the Oracle instance identifier and directory_name denotes a directory that is visible only on the host on which the Oracle instance (SID) is running.
   For example, for an Oracle instance whose SID is Prod01 and an application specific disk /app/prod01/data that is part of the failover package containing the Oracle instance, enter the following value in the Value field:
   Prod01:/app/prod01/data
10. Click Apply Configuration.
    The AuxFailoverMonitor icon, which contains the FOMonitor parameter, is displayed in the PATROL console window, but the InstanceStatus parameter will be highlighted in green on the host on which the instance is running.
    After a failover, the FOMonitor parameter goes into a warning state on each host for sixty seconds. After that, the parameter resets to OK until a failback or another failover occurs.

Debugging failover monitoring

To debug failover monitoring, right-click the Oracle instance icon and choose KM Commands > KM Admin > Debug and set the Debug option to YES.

To debug advanced failover monitoring, activate tracing by using the Activate Tracing menu command available from the AuxFailoverMonitor icon.