Dump processing

Figure 1 shows the example system for dump processing. For this process, the dump volumes are mounted on the cartridge-type devices designated as 3490 with the automatic cartridge loader feature. Because dump processing moves all of the data sets on the volumes that are processed, any data set names on the volumes in each group are not shown.

Assume that it is 3 a.m. on 2003/02/02. This is day 28 of the dump cycle that was established with the DEFINE command. Unless a specific action is taken to prevent it, all hosts that are active begin automatic dump processing for the following reasons:
  • BACKUP in the SETSYS command is specified
  • AUTODUMPSTART(0300 0400 0500) in the SETSYS command is specified
  • It is a Y day in the dump cycle
  • NOREQUEST in the SETSYS command is specified

For SMS active hosts, at the beginning of automatic dump, DFSMShsm obtains volume lists from storage groups defined to automatic dump enabled copy pools and storage groups that are pool type, automatic dump enabled, and have a status of ENABLED, QUIESCED NEW, or DISABLED NEW. If a storage group is both automatic dump enabled and defined in an automatic dump enabled copy pool, it will be processed twice. If a storage group is in more than one copy pool, it can be processed multiple times. Volume priority for dump processing is determined by copy pool association, dump class stacking settings, system affinity, and SMS association. If SMS is not installed or not active, DFSMShsm processes only non-SMS managed DASD volumes.

When DFSMShsm is processing copy pools, volumes are considered candidates if the frequency requirement for the dump class to which the volume belongs has been met.

When DFSMShsm processes volumes within automatic dump enabled storage groups, those volumes are only considered candidates if:
  • The volume has not been processed in any host by automatic dump in the last 14 hours. Any dump created during copy pool processing does not apply to this 14 hour window.
  • The frequency requirement specified for the dump class to which the volume belongs has been met.

A copy pool can be dumped at automatically scheduled intervals with Automatic Dump function. If Auto Dump = Y is specified in the copy pool construct, automatic dump checks generation 0 in the copy pool. If a DASD copy generation 0 exists and has not been previously successfully dumped, automatic dump dumps it. Automatic dump uses the Dump Class specified in the copy pool construct. If the GEN(0) version is partially complete, the failed volumes in the dump classes previously attempted are retried. The automatic dump Y|N settings of the source volumes' storage groups are ignored by automatic dump when it is processing a copy pool. The 14- hours-between-dumps limitation is also ignored by automatic dump when it is processing a copy pool.

DFSMShsm first determines whether there are volumes with dumps that can be stacked. Those volumes associated with the same set of eligible dump classes, of which at least one has a STACK value of more than one, are considered stackable together. Within those volumes, the order is:

  1. Affinity copy pool volumes
  2. Non-affinity copy pool volumes
  3. Affinity SMS volumes
  4. Non-affinity SMS volumes
  5. Non-SMS volumes

Volumes associated with copy pools are not stacked with non-copy pool volumes. Outside of copy pool associated volumes, dumps of SMS-managed and non-SMS-managed volumes can exist on the same dump tape. Affinity and non-affinity volumes cannot be stacked on the same tape. For more information about dump stacking, see DUMPCLASS(STACK/MAXSTACK) and DUMPCLASS(MINSTACK): Specifying the maximum and minimum number of dump copies that are stacked for this dump class.

After all volumes eligible for stacking have been dumped, the remaining volumes are dumped, with DFSMShsm selecting the order of processing as listed above.

After DFSMShsm has attempted to process all the candidate volumes eligible for automatic dump, it retries the selection of any of these volumes found in use by another DFSMShsm function.

Dump processing is performed in three phases in the primary host and in one phase in the nonprimary hosts. These phases are:
  • Delete expired dump copies (primary host only)
  • Dump DFSMShsm-managed volumes for which automatic dump is requested
  • Delete and uncatalog excess dump VTOC copy data sets (primary host only)

If a secondary host is made eligible to take over for a primary host that has failed, it can perform the expiration of dump copies, and deletion of excess dump VTOC copy data sets.