3.3.5 Hot-standby switchover caused by a host pair shutdown

(1) Host pair shutdown for each host in the event of a server failure

If the active server that is ready for hot-standby switchover fails, HA Monitor forcibly terminates the host on which the failure occurred and then performs switchover to the standby host.

Use this function in the following cases:

• Hot-standby switchover is to be promptly performed by forcibly terminating the host without disconnecting individual resources on the host on which the failure occurred.

• Hot-standby switchover is to be promptly performed by forcibly terminating the host without terminating other servers in the same group.

You can use this function on the following servers:

• Servers in the server mode

• Servers in the monitor mode (being monitored by specifying the ptrlcmd_ex operand in the server environment definition)

The active server failures that result in pair shutdown include failures detected by the server itself and server slowdown. If TP1/Server Base or HiRDB is used as a server in the server mode, you can also specify an option that uses only detection of server slowdown as a trigger for pair shutdown.

The active system is reset by the HA Monitor in the standby system that has received a message from the HA Monitor to be terminated. This reset occurs immediately without having to determine the host that has reset priority or the reset issuing host.

Required environment settings

Use the pairdown operand in the server environment definition to specify the pair shutdown function.

To Page Top

(2) Host pair shutdown performed for each host in the event of a user command error

You can also shut down a host pair if a user command that is executed on the source host fails during hot standby processing resulting from a server failure (online -a start) or during planned hot standby processing (online -w start).

Use this function if all the following conditions apply:

• An attempt was made to disconnect the resource of the switching-source host by using a user command.

• After the attempt to disconnect the resource of the switching-source host failed, you want to release the resource by forcibly terminating that host to continue hot-standby switchover.

A user command error means any of the following:

• The user command's termination code is not 0.

• No user commands can be executed.

• A user command was executed, but it does not return control.

  Note that you can handle the preceding errors after the processing times out as specified by using the dev_timelimit or dev_offlimit operand in the server environment definition.

Required environment settings

Specify use for the uoc_pairdown operand in the server environment definition. Check user commands to be executed during hot-standby switchover due to a server failure (online -a start) or during planned hot-standby switchover (online -w start). Make sure that these commands are defined to return a termination code other than 0 in the event of abnormal termination.

To Page Top

(3) Host pair shutdown for each host in the event of a server termination command error

If a server termination command error occurs when hot-standby switchover is performed, you can choose to perform host pair shutdown.

Use this function if you want to forcibly terminate the server and continue hot-standby switchover by forcibly terminating the host after a server termination command error occurred.

A server termination command error means any of the following:

• The server termination command's termination code is not 0.

• The server termination command cannot be executed.

• The server termination command was executed, but it does not return control.

  Note that you can handle the preceding errors after the processing times out as specified by using the stop_timeout or stop_timeout_sw operand in the server environment definition.

Required environment settings

Specify use for the term_pairdown operand in the server environment definition.

To Page Top