Dumping multivolume data sets

An important advantage of DFSMSdss as a backup tool is that it can back up multivolume data sets without having to specify any or all of the input volumes. If you do not specify any input volumes (you are using catalog filtering), multivolume data sets will be automatically processed in their entirety. The catalogs are scanned to select an entire data set; that is, the data set is processed in its entirety from all the volumes it resides on. Logical processing consolidates the extents of the data set in one dump data set for you.

If you specify input volumes using the LOGINDDNAME or LOGINDYNAM volume list, a data set is selected based on the following criteria:

When you either specify SELECTMULTI(ALL) or specify input volumes without specifying the SELECTMULTI keyword, all of the volumes that contain a part of a non-VSAM or VSAM cluster must be in the volume list.
For VSAM data sets, the volume list is affected by the use of the SPHERE keyword as follows:
- Specify SPHERE and you must list all parts of the base cluster in the volume list.
- Do not specify SPHERE and you must list all parts of the base cluster and the associated indexes in the volume list.
When you specify SELECTMULTI(ANY), any part of the non-VSAM data set or VSAM base cluster can be on a volume in the volume list.
For VSAM data sets, the volume list is affected by the use of the SPHERE keyword as follows:
- Specify SPHERE and you must list any part of the base cluster in the volume list.
- Do not specify SPHERE and you must list any part of the base cluster and the associated alternate indexes in the volume list.
When you specify SELECTMULTI(FIRST), the volume list must include the volume that contains the first part of either the non-VSAM data set or the primary data component of the base cluster for a VSAM sphere.
For VSAM data sets, the volume list is affected by the use of the SPHERE keyword as follows:
- Specify SPHERE and you must list the volume that contains the first extent of the data component for the base cluster in the volume list.
- Do not specify SPHERE and you must specify the following information in the volume list:
  - The volume that contains the first extent of the data component for the base cluster.
  - The volume that contains the first extent of the data component for the associated alternate indexes.

Guideline: You are not required to specify the SELECTMULTI option when you build a list of volumes using the STORGRP keyword. The volume list contains all of the volumes in a storage group.

The following is an example of the DUMP command with SELECTMULTI specified:

DUMP  -
    DATASET(INCLUDE(**))  -
    SELECTMULTI  -
    LOGINDYNAM(338001)  -
    OUTDDNAME(TAPE)  -
    COMPRESS

SELECTMULTI works only for logical data set dumps. If you dump a multivolume data set physically, you must ensure that the segments from all the volumes are dumped together. If you dump a multivolume data set physically, it is dumped from all the volumes that are passed. The output dumped data contains a logical file for each selected volume.

A DFSMSdss logical data set dump operation attempts to ensure that all parts of a multivolume non-VSAM data set exist. In cases where a part of the data set is missing, such as an inadvertent scratching of the VTOC entry on a volume, DFSMSdss issues an error message and discontinues processing the data set.

DFSMSdss cannot process the following non-VSAM data sets because they are missing one or more parts:

Multivolume data sets whose catalog volume order differs from the VTOC volume order
Single volume data sets with the same name that are cataloged as one multivolume data set
Multivolume data sets whose last volume indicator in the VTOC entry is not set.

A multivolume data set standard user label is not supported.