The following table describes the processing performed by HiRDB when an RDAREA I/O error occurs.

Table 20-24 HiRDB processing when an RDAREA I/O error occurs

Type of RDAREA	HiRDB processing
	pd_db_io_error_action=dbhold (default)	pd_db_io_error_action=unitdown
Master directory RDAREA	HiRDB (unit for a HiRDB parallel server configuration) terminates abnormally.
Other RDAREAs	The RDAREA is placed in error shutdown status	HiRDB (unit for a HiRDB parallel server configuration) terminates abnormally. The RDAREA is not placed in error-shutdown status, unless the I/O error recurs after the abnormal termination.

The following advantages are provided by configuring HiRDB to terminate abnormally when an RDAREA I/O error occurs:

• If the error is caused by a path error, operations can be resumed upon system switchover.
• The need to recover the RDAREA is eliminated.

(a) If the error is caused by a path error, operations can be resumed upon system switchover

Because HiRDB terminates abnormally when an RDAREA I/O error occurs, the system is switched over. If the cause of the I/O error is a path error, I/O processing can be performed after the system has been switched over, which means operations can resume from that point. As discussed here, path error means that HiRDB cannot access a file because the communication path between HiRDB and the file has been disrupted for some reason.

(b) The need to recover the RDAREA is eliminated

When an RDAREA I/O error occurs, HiRDB terminates abnormally without the RDAREA being placed in error shutdown status. This allows you to eliminate the cause of the I/O error before you restart HiRDB. Once you have eliminated the cause of the error, processing resumes when HiRDB is restarted. The fact that the RDAREA is not placed in error shutdown status in this case eliminates the need for you to recover it, unless, however, the disk needs to be replaced due to a disk error. In such a case, you will need to use the database recover utility to recover the RDAREA.

To configure HiRDB to terminate abnormally when an RDAREA I/O error occurs, you must make the environment settings explained below.

(a) pd_db_io_error_action operand specification

Specify unitdown in the pd_db_io_error_action operand.

(b) If the system switchover facility is being used

If the system switchover facility is being used, see the following sections:

• For standby system switchover (server mode): 26.7.8(2) RDAREA I/O error (path error) has occurred
• For standby-less system switchover (1:1): 26.8.7(2) RDAREA I/O error (path error) has occurred
• For standby-less system switchover (effects distributed): 26.9.7(2) RDAREA I/O error (path error) has occurred

(c) pd_mode_conf operand specification (specify only if the system switchover facility is not being used)

To configure HiRDB so that it will not restart automatically after it terminates abnormally, specify MANUAL2 in the pd_mode_conf operand. If you specify any value other than MANUAL2, HiRDB will restart automatically after it terminates abnormally. When you have specified MANUAL2 and an I/O error occurs in the same RDAREA, that RDAREA will be placed in error shutdown status. This means that the pd_db_io_error_action=unitdown specification is ignored.

(d) Relationships with other functions

If you use functionality provided by the OS or a device driver^# to specify the time until a physical or logical volume I/O error is detected, you must take into account the operands and options described in 8.10 Monitoring the execution time of UAPs and utilities (reducing the effects of nonresponding programs).

If the amount of time specified with the OS or a device driver function is greater than the amount of time specified in one of these operands or options, the target program will be terminated forcibly before the I/O error is detected. To prevent this from happening, you must increase the monitoring times specified with these operands or options.

#: An example of such a function is the -t option of the pvchange or lvchange command in HP-UX. For details about such functions, see the operating system or device driver documentation.

This section explains operations when pd_db_io_error_action=unitdown is specified.

(a) Operations when the system switchover facility is being used

If the system switchover facility is being used, see the following sections:

• For standby system switchover (server mode): 26.7.8(2) RDAREA I/O error (path error) has occurred
• For standby-less system switchover (1:1): 26.8.7(2) RDAREA I/O error (path error) has occurred
• For standby-less system switchover (effects distributed): 26.9.7(2) RDAREA I/O error (path error) has occurred

(b) Operations when the system switchover facility is not being used

If HiRDB terminates abnormally because an I/O error occurred, eliminate the error based on the message that is issued. Restart HiRDB after eliminating the error. Note that if the I/O error recurs after HiRDB is restarted, the RDAREA is placed in error shutdown status. In this case, use the database recovery utility to recover the RDAREA.

(c) Actions to take after HiRDB has terminated abnormally (for both (a) and (b))

If HiRDB terminates abnormally because an I/O error occurred, the pd_db_io_error_action=unitdown specification becomes invalid from that point (RDAREAs will be placed in error shutdown status). To re-enable the pd_db_io_error_action=unitdown specification, use one of the following methods:

• Start HiRDB normally
• Copy the HiRDB system definition files into the $PDDIR/conf/chgconf directory, and then execute the system reconfiguration command (pdchgconf command).

• Because HiRDB terminates abnormally when an I/O error occurs when pd_db_io_error_action=unitdown is specified, if a UAP or a utility is executing in the pre-update log acquisition mode or the no-log mode, the RDAREA being processed might be placed in error shutdown status.
• If an I/O error occurs during startup or termination, HiRDB does not terminate abnormally even if pd_db_io_error_action=unitdown is specified.