HP Serviceguard Toolkit for Oracle Version A.05.01.01 on Linux User Guide (December, 2012)

NOTE:
• This feature is available only on systems running Oracle Clusterware.
• If user defined DB services are configured to start with database package startup, then database
must be started in open mode. Set the package configuration file attribute START_MODE to
open.
The attribute, DB_SERVICE is used to start and stop the database service through the database
package. This attribute is commented by default. You can set the attribute DB_SERVICE to all
or specify the database service names that must be started and stopped by the database package.
To start and stop all the database services configured in the database managed by this package,
set this attribute to all.
For example: tkit/oracle/oracle/DB_SERVICE all
To start and stop one or more specific database services through this database package, you must
specify the service names using multiple instances of this attribute, as shown in the following
example:
Example:
• tkit/oracle/oracle/DB_SERVICE <service1>
• tkit/oracle/oracle/DB_SERVICE <service2>
• tkit/oracle/oracle/DB_SERVICE <service3>
Support for database hang detection
You can configure a separate service that will detect the database hang and take appropriate
action. Database hang is detected by connecting to the database and checking its status. If this
process takes an unusually long time or if the toolkit is unable to retrieve the database status, then
it is assumed that the database is hung. You can configure two attributes when you configure this
service:
1. TIMEOUT: This attribute defines how long the script must wait for the database hang check
to return success. If the database health cannot be determined within this time after connecting
to the database, then it is assumed that the database is hung. It can have any positive integer
as value. The default value for TIMEOUT is 30 (seconds). This must not be confused with the
TIME_OUT package attribute or the service_halt_timeout attribute.
2. ACTION: This attribute defines the action that must be taken if a database hang is detected.
Currently, this attribute can take one of the following values:
• log - Log a message: A message is logged to the package log every time a hang is
detected. If the MONITOR_INTERVAL attribute is set to 30 seconds, then a message is
logged to the package log file every 30 seconds.
• failure - Failover the package to adoptive node.
The default value for ACTION is failure.
The syntax of the service command is as follows:
service_cmd "$SGCONF/scripts/tkit/oracle/tkit_module.sh
oracle_hang_monitor <TIMEOUT> <ACTION>"
Following is an example in which the TIMEOUT is set to 40 seconds and the ACTION is set to
failure:
service_name db_hang_check
service_cmd "$SGCONF/scripts/tkit/oracle/tkit_module.sh oracle_hang_monitor 40 failure"
service_restart none
service_fail_fast_enabled no
service_halt_timeout 300
54 Supporting Oracle features