Out-of-space condition during migration
For each data set that migrates to an ML1 volume and not into an
SDSP data set, DFSMShsm chooses the ML1 volume with the lowest activity
or, as a secondary consideration, the most available space. An out-of-space
condition on the ML1 volumes never causes immediate migration from
an ML1 volume to an ML2 volume. An out-of-space condition occurs when
DFSMShsm:
- Performs migration on a volume and no space exists on any ML1 volume
- Tries to allocate the migration copy and an out-of-space failure occurs
To prevent a large amount of unproductive processing when there
is little free space on any ML1, the migration of data sets—caused
by an out-of-space failure—retries to the ML1 volume having
the most available free space. (The activity to the volume is not
considered.) If a second out-of-space condition occurs while targeting
the ML1 volume that has the most free space, DFSMShsm does the following:
- If the failing data set is small (less than five tracks), and if this is the first time during this processing that a small data set did not fit on an ML1 volume, the task that encountered the out-of-space condition determines the free space on all ML1 volumes, and the migration of the data set is retried. If the task encounters the out-of-space condition again, the free space determination is not done. In this case, the volume migration is terminated and, if ML1 volumes are the only target devices for the volume migration, migration is held. If tape is being used, that part continues.
- If the data set is not small (greater than or equal to five tracks), DFSMShsm fails the migration for the data set but continues migration with the next data set on the source volume.
The previous processing does not occur if the out-of-space condition is caused by a full VTOC for a volume. In this case, DFSMShsm marks the ML1 volume full and tries another.