Before automatic secondary space management starts migration cleanup,
it determines if there is any migration cleanup task running in another
DFSMShsm host. If there is, the migration cleanup will not be performed
on this DFSMShsm host. This avoids any unexpected results that might
be created by running two migration cleanup tasks against the same
MCDS at the same time.
When migration cleanup starts, it checks to see if any internal
TAPECOPY processing is needed, and if so, schedules the internal TAPECOPY
MWEs. Internal TAPECOPY processing could be needed because of duplexing
errors or ABARS recovery of ML2 volumes.
Migration cleanup deletes the following:
- Expired migrated data sets. Migration cleanup deletes the data
sets if:
- For both SMS-managed and non-SMS-managed data sets, the data set
has passed the expiration date indicated in the data set's VTOC
entry and the EXPIREDDATASETS(SCRATCH) parameter is specified in the
SETSYS command.
- For SMS-managed data sets, the data set does not have an expiration
date but has passed the expiration attributes specified in the management
class to which it is associated.
- For an SMS-managed NOSCRATCH GDS, the data set rolled off while
on a level 0 volume and migrated, with rolled off GDG=EXPIRE in the
management class and no expiration date in the VTOC.
See Specifying expiration attributes and Specifying how to scratch expired data sets for details
on SMS-managed data sets.
- Migration copies of the data sets recalled from SDSP data sets.
Because multiple recall tasks can access the small data set packing
data sets concurrently, recall does not erase the migration copy.
Even
though SDSP data sets used during migration cleanup are periodically
released when not in use, and SDSP data sets needed by recall or aggregate
backup are released when needed, it is recommended that you do not
run secondary space management when automatic primary space management
is running in order to avoid contention for SDSP data sets.
Every
30 seconds, migration cleanup checks to determine if recall or aggregate
backup needs an SDSP data set. The value 30 is stored in the MCVT#CHK
field and can be patched to any other desired value.
- MCDS data set (MCD) records for nonreconnectable data sets that
are older than the number of days that are specified by the recalldays value of the SETSYS MIGRATIONCLEANUPDAYS
command.
- MCDS data set (MCD) records for reconnectable data sets that have
passed their predicted remigration date by the number of days that
are specified by the reconnectdays value of
the SETSYS MIGRATIONCLEANUPDAYS command. This occurs only if the recalldays criterion is also met.
- Daily statistics records and volume statistics records that are
older than the number of days specified by the statdays value from the SETSYS MIGRATIONCLEANUPDAYS command.
- Migration copies not scratched during recall, deletion, or GDG
roll-off of non-SMS-managed migrated data sets. These copies exist
if an error occurred during recall or deletion, or if a generation
data set had not passed its expiration date when the roll-off occurred.
Migration
cleanup example
Figure 1 shows an example of SMS-managed volumes
after migration cleanup has occurred for the conditions shown in Figure 1. Data set STANDMC.ML.F2 is deleted from
the ML1 volumes because it has passed the EXPIRE AFTER DAYS NON-USAGE
and EXPIRE AFTER DATE/DAYS values from management class STANDMC. Data
set LARGEMC.ML.G1 is deleted from the ML2 tapes because it has passed
the EXPIRE AFTER DAYS NON-USAGE and the EXPIRE AFTER DATE/DAYS values
from the management class LARGEMC.
Figure 1. Example System of SMS-Managed Volumes after
Migration Cleanup