Nonstop Database, HiRDB Version 9 System Operation Guide

[Contents][Index][Back][Next]

26.3.7 Hitachi HA Toolkit Extension preparations

We recommend that you read this subsection if you are using Hitachi HA Toolkit Extension. This subsection provides guidelines on the values to specify for the HiRDB-related operands in Hitachi HA Toolkit Extension's sysdef and server definition files. For details about how to set up a Hitachi HA Toolkit Extension environment, see the manual Hitachi HA Toolkit.

An operating environment must be set up for each unit.

Organization of this subsection
(1) sysdef definition file
(2) Server definition file

(1) sysdef definition file

(a) servmax operand

You can specify this operand when the version of Hitachi HA Toolkit Extension is 01-04 or later.

For servers that are switched using Hitachi HA Toolkit Extension, specify 16 or 64 as the maximum number of servers that can be started concurrently as running servers or standby servers on a single server machine.

16: Sets 16 as the maximum number of servers that can be started concurrently.

64: Sets 64 as the maximum number of servers that can be started concurrently.

Specify 64 when the number of servers, which are the units for switching on a single server machine, exceeds 16.

The number of servers to be switched using Hitachi HA Toolkit Extension also includes products other than HiRDB that are to be switched. HiRDB computes the number of servers to be switched as follows:

(2) Server definition file

(a) name operand

For the standby system switchover facility, specify the HiRDB directory name using an absolute path name. For a HiRDB parallel server configuration, specify the HiRDB directory name of the applicable unit using an absolute path name.

(b) switchtype operand

Use the switchtype operand to specify the processing to be performed when a server failure is detected.

switch:
When HiRDB (or a unit for a HiRDB parallel server configuration) terminates abnormally, a system switchover is performed and HiRDB restarts on the switchover destination system.

restart:
HiRDB (or a unit for a HiRDB parallel server configuration) restarts in the system where the error occurred. Only if HiRDB cannot be restarted in the system where the error occurred, a system switchover occurs and HiRDB restarts on the switchover destination system (the system switchover is reported by output of the KFPS00715-E message).

manual:
The systems are not switched automatically even if HiRDB (or a unit for a HiRDB parallel server configuration) cannot be restarted. Therefore, do not specify manual in normal cases.
(c) actcommand operand

This operand specifies a command to be executed at the time of package startup or failover. We recommend that you not specify this operand in the server mode, however, because there is no need to issue this command to the standby system unit when a system switchover occurs in the server mode.

When a unit in the system manager is subject to system switchovers, specifying the pdstart command (pdstart -q for a HiRDB parallel server configuration) in the actcommand operand makes it possible to group this command with the package startup command and start HiRDB. However, because the actcommand operand is also executed during system switchovers, the pdstart command is issued to the standby system HiRDB that is already performing the startup processing, and an error is output.

(d) termcommand operand

We recommend that you specify the pdstop -f -q command (pdstop -z -q command for a HiRDB parallel server configuration) for several reasons. First, specifying the forced termination option is guaranteed to stop the unit. Second, during a planned system switchover, the standby system must inherit IP addresses when it restarts. Also, if a planned system switchover is performed when the running system unit has not started, the pdstop command specified in the termcommand operand will result in an error. If the termcommand operand is not specified, the pdstop command must be used to forcibly terminate the running system unit before a system switchover occurs. If a planned system switchover is performed while the running system unit is still active, there will be two running system units. If an error occurs, both systems might stop.