The following table shows the combinations of the versions and products that support data linkage recovery via the system log file.

Table 9-7 Combinations of the versions and products that support data linkage recovery via the system log file

Target system		Source system
		Datareplicator
		08-03 or earlier	08-04 or later
Datareplicator	08-03 or earlier	N	N
	08-04 or later	N	Y
XDM/DS		N	N

Legend:

Y: Supports data linkage recovery via the system log file.

N: Does not support data linkage recovery via the system log file.

The prerequisites for using data linkage recovery via the system log file are as follows:

• The source DBMS is HiRDB.
• HiRDB's system log file has not been overwritten.
• true is specified in the recover_info_send operand in the extraction system definition.
• A recovery information file is available.

The preparations described in this subsection are required to perform data linkage recovery via the system log file.

(a) Creating a recovery information file

When you specify true in the recover_info_send operand in the extraction system definition, a recovery information file is created under the target Datareplicator's $HDSPATH directory. Recovery information will be stored in this file. This information is required to determine the input start position in the system log file during recovery processing. The following table describes the contents the recovery information file.

Table 9-8 Contents of the recovery information file

Item	Contents	Remarks
File name	rcvrfile_xx_yy	xx: Target Datareplicator identifier (hdsid) yy: Data linkage identifier (dsidxxx) Specify both as two hexadecimal characters. For alphabetical characters, use lower-case letters.
Location of file creation	Under the target Datareplicator's $HDSPATH	--
Unit of creation	One file per target Datareplicator data linkage identifier (dsidxxx)	--
File type	Regular file or character special file	If the target Datareplicator is in a system switchover environment, use a character special file.
File size	(1 + n) KB	The size depends on the import method: Transaction-based import method n: 1 Table-based import method n: Number of import groups Recovery information is stored for each import group.
Creation timing	When port check information is received	--
Update timing	When an import synchronization point is obtained (when the import status file is updated)	Update information is overwritten, not accumulated.

Legend:

--: Not applicable

Transmission of recovery information

Because the recovery information is sent to the target Datareplicator for each transaction that occurs during extraction processing, the amount of transmission data and the usage amount of the target Datareplicator's import information queue file increase during normal operation. The formula for determining the amount of increase in the resources needed to send recovery information is shown below. When you design the resources, take into account this increase.

• Amount of transmission data
  

  (TRUN_NUM / SND_INT) x 144 (bytes)

• Size of the import information queue file
  

  (TRUN_NUM / SND_INT) x 144 (bytes)

• Size of the extraction information queue file
  

  144 x TRN_NUM (bytes)

TRN_NUM: Number of transactions that updated the table subject to extraction processing

SND_INT: recover_info_send_interval operand value (1 if omitted)

Recovery information transmission interval

To minimize the amount of increase in transmission data during normal operation, be sure to specify a recovery information transmission interval in the recover_info_send_interval operand in the extraction system definition.

We recommend that the value you specify in this operand be based on the average number of transactions per transmission interval, in the range from 1 to 32,767, so that recovery information will be added at least once per transmission interval (sendintvl value specified in the transmission environment definition).

The following explains what happens if you specify a value other than 1 in the recover_info_send_interval operand:

After a transmission process starts during normal operation, Datareplicator adds recovery information to the transmission data for the first transaction to be sent. Thereafter, Datareplicator sends recovery information every specified number of transactions. In this case, during recovery processing, Datareplicator will input and extract update information including for instances that do not require recovery.

The following figure shows an example of recovery information transmission when 5 is specified in the recover_info_send_interval operand.

Figure 9-4 Example of recovery information transmission

Explanation:

The update information that needs recovery due to an error is Update 11 through Update 5. Because the recover_info_send_interval operand is specified, Datareplicator inputs and extracts the update information starting at Update 6. The update information that is not needed for recovery will be input and extracted, but not sent.

Note:

Recovery processing cannot be performed, even if update information whose recovery is not needed has been overwritten.

Regardless of this operand's value, Datareplicator sends the recovery information during event transmission. The following figure shows an example of recovery information transmission during event transmission when 5 is specified in the recover_info_send_interval operand.

Figure 9-5 Example of recovery information transmission during event transmission

(b) Checking the capacity of the system log file

If a value other than 1 is specified in the recover_info_send_interval operand in the extraction system definition, allocate a sufficient capacity to the system log file so that no system log information that will be required for recovery will be overwritten by the system log during the period from occurrence of an error to the completion of recovery.