When the source HiRDB is a parallel server and mutual system switchover configuration is used, system switchover might result in multiple active node master processes, all of which belong to the same source HiRDB on a single server machine. If this happens, each node master process's directory is duplicated on the same server machine; as a result, filenames under the operation directory might be duplicated among the node master processes.

Figure 6-31 shows an example of a mutual system switchover configuration when the source HiRDB is a parallel server, and Table 6-22 shows the duplicate files in the operation directory based on server machine 3 in Figure 6-31.

Figure 6-31 Example of mutual system switchover configuration when the source HiRDB is a parallel server

Table 6-22 Duplicate files in the operation directory based on server machine 3

No.	Active server			Files under Datareplicator directory
	MGR	BES1	BES2
1	--	--	Y	Queue files Status files Data linkage file Error information files Activity trace files
2	Y	--	Y	Queue files Status files Data linkage file Error information files Activity trace files
3	--	Y	Y	Master status file Master error information files Queue files Status files Data linkage file Error information files Activity trace files
4	Y	Y	Y	Master status file Master error information files Queue files Status files Data linkage file Error information files Activity trace files

Y: Active server

--: Inactive server

___ (underline): Files duplicated between node master processes

If multiple node master processes are active, the error information and activity trace files might be duplicated. When this type of system switchover configuration is used, specify true in the errfile_unique operand in the extraction system definition in order to prevent duplication of the error information files and the activity trace files. If you specify false in the errfile_unique operand and system switchover occurs with this organization, restart of the node master process will result in an error at the standby system.

To monitor the Datareplicator's status, monitor the messages that are issued by the Datareplicator.

During system switchover, the cluster software controls activation and deactivation of the disk. However, if the disk is to be initialized after system switchover is stopped, the user must activate the disk manually. Before initializing the disk, make sure that you activate the disk. The disk cannot be initialized unless it is activated and made accessible prior to initialization.

In a hot-standby configuration, synchronize the time settings in the running machine and the standby machine. If their times are not synchronized when system switchover occurs, the extraction process cannot extract update information from the system log file.