R+/CHANGE ACCUM phases

ANALYZE phase

During the ANALYZE phase, the R+/CHANGE ACCUM product:

• Parses all commands in SYSIN

• Retrieves the information about change accumulation group from the repository

• Reads the Db2 catalog and directory to retrieve object descriptions, image copy LRSNs, and Db2 log range data

• Selects the log point at which to end extraction

  This log point is the highest LRSN written to the Db2 log, as of a point in time during the ANALYZE phase.

• Selects the appropriate change accumulation files or Db2 log files to read for extraction of the necessary log data

• Eliminates spaces that are unrecoverable from the run, even though execution will continue as long as at least one space is recoverable in the group

  An example of an unrecoverable space is one that has a LOAD LOG(NO) entry in SYSCOPY that is not followed by an image copy. An R+/CHANGE ACCUM run with an unrecoverable space ends with return code 12.

LOG INPUT phase

During the LOG INPUT phase, R+/CHANGE ACCUM extracts any appropriate Db2 log records for all target objects and passes them to a sort. R+/CHANGE ACCUM extracts log records for each target object for a range beginning at the last change accumulation file that contains records for the object, or at the last image copy for the object if there are no usable change accumulation files. The range extends to the log LRSN that R+/CHANGE ACCUM selected during the ANALYZE phase.

MERGE phase

R+/CHANGE ACCUM does the actual work of creating new change accumulation files in one or more MERGE phases.

R+/CHANGE ACCUM executes one MERGE phase for each of the following objects:

• Each partition of a partitioned space

• A nonpartitioned space for which you specified DSNUM ALL

• DSNUM n of a nonpartitioned space (if you specified DSNUM n in the definition in the change accumulation group)

Each MERGE phase may read parts or all of one input change accumulation file. The phase merges the appropriate records with sorted Db2 log records to write output records to an output change accumulation file. If a table space is included in more than one group for which you specified ACCUM GROUP statements, the MERGE phase for the table space, and its associated indexes (if requested), will write output records to more than one set of files. More than one duplicate file may be written for each group. If a group includes more than one target object, several MERGE phases will write records to the output change accumulation file for the group.

Output change accumulation file sets are not recorded in the repository until the last MERGE phase has completed writing records to its files. R+/CHANGE ACCUM records the appropriate change accumulation log ranges in the repository. For a partitioned space, R+/CHANGE ACCUM always records the log ranges at the partition level, even if the definition in the group specified DSNUM ALL. However, for nonpartitioned, multi-data-set spaces, R+/CHANGE ACCUM records log ranges at the same level as the definitions. If you specified DSNUM ALL, R+/CHANGE ACCUM records one log range for the table space. If you specified DSNUM n, the MERGE phase for DSNUM n records one log range.