To minimize the time required to perform a system switchover, the standby-less system switchover (effects distributed) facility opens only the RDAREAs needed for full recovery when a system switchover occurs. It does not open other RDAREAs. The RDAREA opening trigger attributes are as follows:

• Even if N is specified in the pd_rdarea_open_attribute_use operand, Y is assumed.
• When a system switchover occurs, even if the RDAREA opening trigger attribute is INITIAL, DEFER is assumed.

For details about the RDAREA opening trigger, see the explanations for the pd_rdarea_open_attribute_use and pd_rdarea_open_attribute operands in the manual HiRDB Version 9 System Definition.

When the situation described below occurs in an environment in which a global buffer for OTHER is defined, you must not attempt to terminate the unit alone by executing the pdstop -u command from the unit where the system manager is defined. First, either execute the pdstop command from the unit where the system manager is defined to terminate HiRDB, or use the system reconfiguration command (pdchgconf command) to allocate a global buffer to the RDAREA to be added. Then, terminate the unit. For details about global buffers for OTHER, see 26.5.2(5) Defining global buffers (standby-less system switchover (effects distributed) facility).

• In a unit to which the standby-less system switchover (effects distributed) facility is applicable, the globalbuffer operand was omitted from the create rdarea statement of the database structure modification utility (pdmod), and an RDAREA with a page size larger than the size of the global buffer for OTHER was added.

If you change a HiRDB system definition or the HiRDB configuration, do not use a cluster software command to terminate HiRDB. Instead, use the pdstop command to normally terminate HiRDB only, and then change the HiRDB system definition or reconfigure HiRDB. When you are done, you can use the pdstart command to start HiRDB only.

If HiRDB is stopped by the cluster software, it might not be possible to manipulate HiRDB files on the shared disk from either system. In this case, you must use an OS command to activate the shared disk.

If you are using Hitachi HA Toolkit Extension, the standby system unit will not wait for the running system unit to start. If a command is issued to the running system unit before the standby system unit has started, an error might occur and the standby system unit might abort (Phi1012). Therefore, wait for startup of the running system unit to finish before starting the standby system unit. If the standby system unit is started before startup of the running system finishes, the standby system unit might abort (Phi1012).

When a system switchover occurs, the client connections terminate and the database returns to its status before the transactions began. If necessary, re-execute the processing.

You must be careful about executing the pdsetup -d command when HiRDB is forcibly or abnormally terminated. Do not enter Y at the command's prompt. If you do enter Y, it might become impossible to start HiRDB subsequently.

Note the following points about using HA Monitor:

• During startup of a server, if all systems, including the primary system (system for which online is specified in the init operand of HA Monitor's servers definition), wait for the running system server to start,^#1 start the primary system^#2 as the running system by executing HA Monitor's monact command.
• If a server is being activated in a unit to which the standby-less system switchover (effects distributed) facility is applied, do not forcibly terminate that unit. If it is forcibly terminated, the next time a server is started on the primary system, this server might wait for the running system server to start.^#1 If this happens, start the primary system^#2 as the running system by executing HA Monitor's monact command.

#1

Status in which *SBY* is displayed when HA Monitor's monshow command is executed.

#2

If the secondary system (system for which standby is specified in the init operand of HA Monitor's servers definition) server is started by the monact command as the running system, the messages listed below might be output repeatedly on the primary system until startup of the running system is complete. These messages cease to be output once startup of the running system is complete.

• KFPS05608-I
• KAMN305-E
• KAMN222-I

The KAMN305-E and KAMN222-I messages are output by HA Monitor.

If the time used by the OS differs significantly between the primary system machine and the standby system machine (that is, if the time delay on the standby system machine is greater than the time required to perform a system switchover), note that HiRDB will be affected in the same way as when an incorrect operation is performed in an attempt to move the time back.

For details about how to change the time used by the OS, see 9.12 Changing the time used by the OS.