Using VOL=SER within a job step

When data set stacking is specified using VOL=SER within a job step, the system ensures that all the data sets making up the data set collection (a group of data sets intended to be allocated on the same tape volume or set of tape volumes as a result of data set stacking) are directed to the same device category. If the ACS routines initially directed the stacked allocations to different device categories, the system detects this and reinvokes the ACS routines, passing additional information to those routines. The ACS routines can then do one of the following:

Correct the problem and route the allocations to consistent device categories
Fail the stacked allocation (if the ACS routine exits with a nonzero return code)
Fail to correct the inconsistency, in which case SMS fails the allocation

Recommendation: The system cannot detect data set stacking if VOL=SER is used to stack data sets across jobs or job steps. In these instances, you can change your JCL to specify VOL=REF instead of VOL=SER. For more recommendations on when to use VOL=REF versus VOL=SER for data set stacking, see z/OS MVS JCL User's Guide.

The system reinvokes the ACS routines only when all of the following conditions are true:

The request is part of a data set collection based on a data-set-sequence-number greater than one specified on the LABEL parameter, and a VOL=SER, where at least one of the volume serial numbers matches one of the volume serial numbers for a previous request in the same step.
The request is currently directed to a different device category than the other requests in the data set collection.
The request is DISP=NEW (or DISP=MOD treated as NEW).