Planned system switchover

This subsection explains how to perform a planned system switchover.

The two methods described below are available for planned system switchovers. Select the one that best suits the environment you are using.

Method 1
With this method, planned system switchovers are executed after transaction queuing is performed. In this case, new transactions are queued even during a planned system switchover. Therefore, executing a UAP during a planned system switchover does not result in an error.
This method can only be used for planned system switchovers if queuing is specified in the pd_ha_transanction operand.
When this method is used for a planned system switchover, UAP transactions are queued and placed in waiting status after the pdtrnqing command is executed and until the planned system switchover is complete. This increases the response time. Therefore, check whether the following setting is appropriate:
- Client maximum wait time specified in the PDCWAITTIME operand of the client environment definition
When the time specified by this operand has passed, the UAP ends in an error.
Method 2
With this method, planned system switchovers are executed without using transaction queuing. In this case, new transactions are not queued during a planned system switchover. Therefore, executing a UAP during a planned system switchover results in an error.

Organization of this subsection: (1) Using method 1 to execute a planned system switchover; (2) Using method 2 to execute a planned system switchover

(1) Using method 1 to execute a planned system switchover

The planned system switchover procedure differs depending on the cluster software product that you use.

(a) When the cluster software product being used is HA Monitor

The planned system switchover procedure is as follows.

Procedure

Use the pdls -d ha -s server-name -a command to confirm the following:
The source unit of the back-end server subject to the planned system switchover is active.
The switchover target unit of the back-end server subject to the planned system switchover is in standby status.
Execute the pdtrnqing command at the source unit and start transaction queuing.
During this step, confirm that the pdtrnqing command terminated normally with a return code of 0. If the return code is 0, transaction queuing has started normally. If the return code is not 0, transaction queuing has not started. In this case, eliminate the cause of the problem, and then re-execute the pdtrnqing command.
If the switchover target is a normal unit, use the monswap command of HA Monitor to perform the planned system switchover. Afterwards, proceed to step 8.
If the switchover target is an accepting unit, proceed to the next step.

Reference note

If there are multiple accepting units, you cannot use the monswap command of HA Monitor to specify the switchover target. Switchover targets are determined according to the priorities that were specified during system configuration. Therefore, if the switchover target selected according to priority is the intended unit, use the monswap command to execute the planned system switchover. If the switchover target is not the intended unit, follow steps 4 through 7 to execute the planned system switchover.
From the unit where the system manager is located, execute the pdstop -f -s server-name command to forcibly terminate the back-end server. During this step, confirm that the pdstop command terminated normally with a return code of 0.
From the unit where the system manager is located, execute the pdstart -s server-name -u switchover-target-unit-identifier command. During this step, confirm that the pdstop command terminated normally with a return code of 0.
Use the pdls -d ha -s server-name -a command to check the status of the back-end server at the switchover target unit.
If the back-end server is waiting for the running system to start at the switchover target unit, proceed to the next step.
If the back-end server is active at the switchover target unit, proceed to step 8.
Execute the monact command of HA Monitor at the switchover target unit. During this step, confirm that the monact command terminated normally with a return code of 0.
Use the pdls -d ha -s server-name -a command to confirm that the back-end server is active on the switchover target unit.

When the planned system switchover is complete, transaction queuing is automatically canceled, and the transactions that have been queued are executed. If transaction queuing is not canceled for some reason, execute the pdtrnqing -d -s server-name command on the switchover target unit to cancel transaction queuing. During this step, confirm that the pdtrnqing command terminated normally with a return code of 0.

(b) When the cluster software product being used is not HA Monitor

The planned system switchover procedure is as follows.

Procedure

Use the pdls -d ha -s server-name -a command to confirm the following:
The source unit of the back-end server subject to the planned system switchover is active.
The switchover target unit of the back-end server subject to the planned system switchover is in standby status.
Execute the pdtrnqing command at the source unit and start transaction queuing.
During this step, confirm that the pdtrnqing command terminated normally with a return code of 0. If the return code is 0, transaction queuing has started normally. If the return code is not 0, transaction queuing has not started. In this case, eliminate the cause of the problem, and then re-execute the pdtrnqing command.
Use a cluster software command to execute a planned system switchover.
Use the pdls -d ha -s server-name -a command to confirm that the back-end server is active on the switchover target unit.

(2) Using method 2 to execute a planned system switchover

For a planned system switchover, execute HA Monitor's monswap command.

When this command executes, the back-end server at the switching-source unit is automatically placed on standby status.

If you are using Hitachi HA Toolkit Extension, enter the cluster software's planned system switchover command instead of the monswap command.

The following table describes the planned system switchover operation when the standby-less system switchover (effects distributed) facility is used.

Table 26-74 Planned system switchover operation when the standby-less system switchover (effects distributed) facility is used

Back-end server type	Operation	Move destination
Host BES	Moves to the switching destination with the highest priority and, if there is no free space in the guest area, it moves to the switching destination with the next highest priority.	Guest BES
Guest BES		Host BES

(a) Planned system switchover for a host BES

The figure below shows an example of a planned system switchover for a host BES. In this example, a server machine (hostA) must be stopped for hardware maintenance of the host.

Figure 26-117 Example of a planned system switchover for a host BES

[Figure]

In this example, a planned system switchover is executed for the current BES on unt1. The procedure is as follows:

At hostA, enter the monswap command for BES1 to switch BES1 to hostB.
At hostA, enter the monswap command for BES2 to switch BES2 to hostC.

If you are using Hitachi HA Toolkit Extension, enter the cluster software's planned system switchover command.

(b) Planned system switchover for guest BESs (system reactivation)

The figure below shows an example of a planned system switchover for guest BESs (system reactivation). In this example, a unit recovers from an error.

Figure 26-118 Example of a planned system switchover for guest BESs (system reactivation)