The Makers of FDR®

July 31, 2006

Innovation Data Processing Announcements

IMPORTANT NOTICE FOR ALL FDRABR CUSTOMERS ON Z/OS 1.7

Innovation has discovered that an IBM change in z/OS 1.7 has affected the ability of FDRABR to keep track of backups of sequential (DSORG=PS) and partitioned (DSORG=PO) data sets; this problem also affects z/OS 1.8.

All z/OS 1.7 ABR customers should install the fixes for IBM APARs OA17290, OA17365 and OA17370. If you are migrating to 1.7 or 1.8, insure that these fixes are installed first.


z/OS 1.7 introduced “large sequential” data sets, DSORG=PS data sets over 64K tracks in size. These data sets are created using the DSNTYPE=LARGE parameter in JCL or in the data class. There is a flag in the F1 DSCB (DS1LARGE) indicating a large data set.

For large sequential data sets, IBM claimed a byte in the F1 DSCB which was formerly reserved and used it for an extension of the “last block pointer” so that it could point to more than 64K tracks. The byte is now called DS1TTTHI. This byte is also used by FDRABR to keep track of the backups of individual data sets.

FDRABR V5.4 level 50 was modified so that it no longer uses that reserved byte, but ONLY if the DS1LARGE flag is on. IBM assured us that no IBM code would use or modify DS1TTTHI unless the DS1LARGE flag was on. This allowed us to maintain compatibility with existing data sets and backups, changing our operation only for the large sequential data sets.

However, several IBM bugs have been found where DS1TTTHI is incorrectly zeroed in existing DSORG=PS/PO data sets even when DS1LARGE is off. If this occurs after the data set has been backed up by FDRABR (DUMP TYPE=FDR, TYPE=ABR, or TYPE=AUTO), then the records of the backup for this data set will be lost.

This does not affect backups taken with PGM=FDR or FDRDSF, nor does it affect archive (DUMP TYPE=ARC) or application backup (DUMP TYPE=APPL).

Note: the backups themselves are not harmed. You can still restore data sets and entire volumes from the backups. Only the ability to restore a data set by data set name without specifying GEN= and CYCLE= is lost. The next backup taken after the IBM fixes are applied will be recorded properly.

  • OA17290 – OPEN/CLOSE – if an existing DSORG=PS/PO non-large data set is opened for output, DS1TTTHI is zeroed. The PTFs for this APAR are UA28030 (1.7) and UA28031 (1.8)
  • OA17365 – CHECKPOINT/RESTART – if a DSORG=PS non-large data set is opened for output during restart, DS1TTTHI is zeroed
  • OA17370 – DFSMSdss – some DSS (ADRDSSU) functions may zero DS1TTTHI on a non-large DSORG=PS data set
All FDRABR volume backup users running or installing z/OS 1.7 should apply these fixes. If the final PTFs are not yet available, ++APAR fixes may be available from IBM level 2. ABR customers installing z/OS 1.8 later this year should verify that these fixes are applied, and apply them if not.