The following table describes the system processing and the actions that the HiRDB administrator takes when an error occurs.

Table 26-32 System processing and actions the HiRDB administrator takes when an error occurs (when using the system switchover facility)

Error	System processing	Action the HiRDB administrator takes
A server failure occurred (only HiRDB terminated abnormally).	Does not perform a system switchover. Restarts HiRDB in the system where the error occurred.	Check the cause of the error based on the message that was output. For details about the appropriate action to take, see Chapter 20. Error Handling Procedures.
A system failure occurred.	Performs a system switchover and restarts HiRDB in the switchover destination system. However, depending on the type of system failure, a system switchover might not be performed. A cluster software specification determines whether a system switchover will be performed. For details, see the cluster software documentation.	Check the cause of the error based on the message that was output. For details about the appropriate action to take, see Chapter 20. Error Handling Procedures or the applicable manual.
HiRDB restart failed.	As specified in the pd_ha_restart_failure operand#.
System switchover failed.	Terminates the system.

Note

The information inherited after a system switchover is the same as the information that is inherited when HiRDB is restarted. For details about the information that is inherited during a HiRDB restart, see 20.1.5 Information inherited during a HiRDB restart.

#

If you register a command for switching the cluster software system in this operand, the system will be switched automatically when restart processing fails. However, the system stops operating if this operand is omitted. In this case, you must enter a command for switching the cluster software system, and then perform a system switchover.

(a) Actions to take when a system switchover occurs

When a system switchover occurs, VERITAS Cluster Server takes all resources on the running system offline, starting with the parent resource, and then places the resources on the standby system online starting with the child resources. If an unexpected system switchover occurs, take the following actions as necessary to return the running system to its original server machine:

• If a process on the running system server machine cannot continue
  Determine the reason why the server machine process cannot continue, and eliminate the error. Then, activate VERITAS Cluster Server.
  
• If an error occurs in a resource
  Determine the cause of the resource error, and eliminate the error. Then, use a VERITAS Cluster Server command to release the resource's error status.
  

(b) Action to take when HiRDB terminates abnormally

When HiRDB terminates abnormally, restart HiRDB on the running system. When a dummy file for a HiRDB_x type resource is deleted, the resource is placed in error status, and HiRDB is forcibly terminated. In this case, perform the following procedure to restart HiRDB.

Procedure

1. Use a VERITAS Cluster Server command to release the error status of the resource.
2. Use a VERITAS Cluster Server command to bring the resource online. HiRDB then restarts.