BY-PART—Process Objects By Partitions
The BY-PART keyword specifies whether to process each tablespace partition separately.
The BY-PART keyword specifies whether to process each tablespace partition separately for the specified database. When you use BY-PART, you do not need to specify each partition individually.
BY-PART is valid only with the RECOVER DATABASE statement.
If any of the tablespaces in the database were previously copied at the partition level (one copy per partition), BY-PART is required.
If you do not specify BY-PART in the SYSIN, the BY-PART value from the PFR parmlib member (
hlq.CDBAPARM) is used.
This keyword has the following format:
- YESProcesses each partition separately. For more information, see Recover a Partitioned Tablespace or Indexspace.BY-PART YES is the same as using DSNUM to specify each partition, or using LISTDEF with the PARTLEVEL option.
- NOProcesses the database as a whole, not by partitions. This value is the default.If ALL-PARTS YES is specified in the PFR parmlib member, it overrides BY-PART NO in the SYSIN. Each partition is processed separately.If objects are specified using wildcards, they are always processed by partition.
Example: Recover Database Tablespaces by Partition
This example recovers tablespaces in the DB1 database by partition:
RECOVER DATABASE DB1 BY-PART YES
Example: Use Multiple BY-PART Keywords in a Job
This example combines processing of tablespaces by partitions and as a whole in one RECOVER statement. This example assumes that ALLPARTS NO and BY-PART NO is specified in the PFR parmlib member. The tablespaces in databases DB1 and DB3 are processed by partition (BY-PART YES keyword). The other databases are processed as a whole (BY-PART NO in the parmlib member).
RECOVER DATABASE DB1 BY-PART YES DATABASE DB2 DATABASE DB3 BY-PART YES DATABASE DB4