Processing Logic

By default, the archive data sets are sorted in alphanumeric order that is based on the volume serials. You can modify the sysparm MERGSORT to sort the data sets by least space that is used or custom user sort. All of the input selection tests are applied to determine which archive volumes (tapes or disk archive data sets) to process.

The ARCHVOLS record corresponding to each of the selected volumes is then examined (a flag bit is tested) to ensure that the volume closed properly. If the volume did not close properly, it can be in use. Do not process the volume by merge. This condition can occur if archive and merge jobs are running simultaneously. To maintain data integrity, even if the jobs are running concurrently over midnight, CA Disk merges an open volume only if its creation date is less than today's date minus one (CREDT < TODAY-1).

The file number (position) sorts the DSNINDEX records within each archive volume. This permits all data sets being merged to be located by skipping forward, whether this is a tape spin or a seek of the disk heads. Each unexpired data set is merged forward to a new archive volume. The data set index entry is updated to reflect its new location (live mode only). The data set expiration date with the LIMITS= parameter determines which of the new output devices is used. If its expiration date is greater than the highest limit specified, it is merged to the archive data set associated with the highest limit.

Note: Since SMS-managed DSNINDEX records have an expiration date of 1999.365, they are always written to the ARCHVOL associated with the highest limit.

Index entries for expired data sets are deleted. When all index entries for a volume have been either deleted or updated to reflect their new locations on a new archive volume, the old ARCHVOLS record is deleted. When any of the moved data sets are to be restored, CA Disk calls for the new archvol.

Each archive volume expiration date that merge creates is taken from the sysparm value for DYNEXPDT. Also, the date can come from the value you specified in an overriding JCL. The default value of DYNEXPDT is the never expire Julian date of 99365. If you specify a different non-zero value, it is used. In an EDM environment, Leave DYNEXPDT at its default value, which allows DSNINDEX and ARCHVOL records to expire correctly.

For more information, see the section Year 2000 Considerations in the chapter "General Information."

To determine the average length of archive tape volumes being used, merge processing relies on the specification of sysparms. The information determines the amount of space that is used on each ARCHVOL. Be sure to review sysparms CARTCALC, CARTFEET, CARTPRCT, and TAPEFEET, of the Systems Guide, and include them in the SYSPARMS member of parmlib if needed.

Note: The data set index entry expiration date determines candidates to merge forward. This means that if the EXPDT or RETPD parameters were hard-coded on the //ARCHIVE0 DD statement during an archive run, you can have an unexpired data set on an expired archive volume. The unexpired data set is still called for in merge processing to ensure the integrity of merge. No unexpired data set is dropped. If the expired tape has been written over or the archive data set on disk has been deleted, errors result when trying to locate the data set to be merged. When you use hard-coded JCL expiration dates, run the VOLDELETE command of IXMAINT before running merge to prevent the previous problem.

Merge processing creates report records for all volumes and data sets processed, indicating the actions that are taken. They are created concurrently with the MERGE and INDEX UPDATE step, but must be resorted into the proper sequence for the report writer.