(a) Starting OpenTP1

By using the OpenTP1 start command dcstart, you can start two OpenTP1 systems that use the System Switchover facility. If OpenTP1 is set to start automatically, you do not need to use dcstart.

• The system environment definition (i.e., whether or not manual is specified) and the type of recent termination determines whether OpenTP1 is started automatically or manually.
• The running or standby system can be specified in the server statement that sets the HAmonitor environment.
  A system for which online is specified in the initial operand or for which the initial operand is omitted becomes the running system when online operation starts.
  A system for which standby is specified in the initial operand becomes the standby system when online operation starts.
  
• Whether OpenTP1 is started normally or restarted is decided in the same way as when the System Switchover facility is not used.

Having the standby system perform the postprocessing of the running system after a system switchover

To have the standby system perform the postprocessing of the running system after a system switchover, execute dcstart -U to start the standby system.

When dcstart -U is used to start the standby system, the user server does not start at a system switchover. The standby system started in this way can be used for only the postprocessing of the running system, such as deciding undecided transactions and ensuring database integrity. The machine for operating the standby system in this case does not require resources equivalent to those for the running system.

Because the dcstart command needs to be executed, the standby system when used for postprocessing must be in a state in which it can be started manually.

The following describes an operation when one node is used as a standby system to perform only the postprocessing of multiple OpenTP1 systems.

In this configuration, a standby OpenTP1 system must be set up for each running OpenTP1 system.

Figure 6-3 Operation in which a standby system performs only the postprocessing of a running system after a system switchover

Explanation:

1. An error occurs in the OpenTP1 D running system.
2. The OpenTP1 D' standby system starts.
   OpenTP1 D' performs postprocessing, such as deciding undecided transactions and ensuring database integrity. Meanwhile, OpenTP1 A, OpenTP1 B, and OpenTP1 C continue online processing in fall-back mode. Because the number of running systems is reduced from 4 to 3, the system performance is degraded by 25%.
   

Notes:

In the following cases, a standby system does not start for the sole purpose of postprocessing:

• When the mode_conf operand (system startup method specification) in the system environment definition has been set to AUTO, and OpenTP1 terminates abnormally or the OS is started in a situation in which the previous OS termination was not abnormal
• When the mode_conf operand (system startup method specification) in the system environment definition has been set to MANUAL1, and OpenTP1 terminates abnormally

Even when the operation has been set so that the standby system will perform only the postprocessing of the running system at a system switchover, if one of the conditions above is satisfied, the dcstart command is automatically executed. The standby system started by the dcstart command without an option waits to take over from the running system.

If you want to start the standby system again for the sole purpose of postprocessing, stop OpenTP1 and then restart it by using the dcstart -U command.

(b) Terminating OpenTP1

Tables 6-1 and 6-2 describe how to terminate an OpenTP1 system that uses the System Switchover facility.

Table 6-1 Commands for terminating a running OpenTP1 system

Termination commands	Execution results for a running OpenTP1 system
dcstop (with no option)	Terminates the running system normally. The standby system also terminates.
dcstop (with -n option)	Forces the running system to terminate normally. The standby system also terminates.
dcstop (with -a option)	Terminates the running system in the planned termination A mode. The standby system also terminates.
dcstop (with -b option)	Terminates the running system in the planned termination B mode. The standby system also terminates.
dcstop (with -f option)	Terminates the running system forcibly. The standby system also terminates.
monswap (HAmonitor command for a planned system switch)	The systems switch after the running server terminates.
monsbystp (HAmonitor command for terminating the standby system)	Cannot be executed.

Table 6-2 Commands for terminating a standby OpenTP1 system

Termination commands	Execution results for a standby OpenTP1 system
dcstop (with no option)	Cannot be executed.
dcstop (with -n option)	Cannot be executed.
dcstop (with -a option)	Cannot be executed.
dcstop (with -b option)	Cannot be executed.
dcstop (with -f option)	Terminates the standby system forcibly.
monswap (HAmonitor command for a planned system switch)	Cannot be executed.
monsbystp (HAmonitor command for terminating the standby system)	Terminates the standby system.

(a) Note on commands executed offline

When you execute commands offline, you must first terminate both OpenTP1 systems in a system switch configuration, and then execute the commands. You can execute the dcstart command if an OpenTP1 system which you want to start is not started. The other OpenTP1 system need not be terminated. A standby system ignores the dcstart -n command.

(b) Note on commands executed online

In a running OpenTP1 system, you can execute commands online. In a standby OpenTP1 system, you cannot execute any commands online except for the dcstop -f command (for a forced termination).

To reduce the burden of unloading the running system, system journal files for the standby system can be unloaded. Be careful, however, when executing the jnlunlfg command for an unload. For details of unloading the standby system, see the description of the jnlunlfg command in the manual OpenTP1 Operation.

(c) Accessing a shared disk device

Even if you execute a command to perform an operation on a shared file on a shared disk device, OpenTP1 does not write to the file so the file will not be destroyed. Note, however, that if you execute the current system only (with the secondary system usable but not yet activated as an OpenTP1 component), you can write to a shared file from the secondary system and using a command in this case might destroy files. You must terminate both systems before using a command to perform operations on a character special file of OpenTP1.