For more information about allocation and space requirements, see the CICS System Definition Guide.) For extrapartition transient data considerations, see “Recovery for extrapartition transient data” on page 134.

You must specify the name of every intrapartition transient data queue that you want to be recoverable in the queue definition. The recovery attributes you can specify for an intrapartition transient data queue are:

vLogical

vPhysical

vNone

Logical recovery

If you request logical recovery on an intrapartition queue definition, changes to a transient data queue by an interrupted unit of work are backed out. Backout occurs dynamically in the case of a task abend, or at a CICS emergency restart in the case of a CICS failure.

As a general rule, you should request logical recoverability. For example, if you make related changes to a set of resources that includes intrapartition transient data, and you want to commit (or back out) all the changes, you require logical recovery.

Physical recovery

Physical recoverability is unique to transient data and is effective on both warm and emergency restarts. By requesting physical recovery on an intrapartition queue definition, you ensure that changes to the queue are committed immediately and, with one exception, are not backed out.

The exception is in the case of the last read from a physically recoverable queue before a unit of work fails. CICS always backs out the last read from a physically recoverable transient data queue. In terms of the read and write pointers that CICS maintains for TD queues, this means that the read pointer is reset, but the write pointer never changes. This is illustrated by the diagram in Figure 12 on page 133. The sequence of TD actions in this example, and the subsequent recovery, is as follows:

vThe unit of work has read items 1 and 2, leaving the read pointer at item 3.

vThe unit of work has written item 4, leaving the write pointer ready for the next item to be written.

vCICS abends and is restarted with an emergency restart.

vAs a result of the transient data recovery, the read pointer is reset to item 2, queue items 2, 3, and 4 are still available, and the write pointer is restored.

132CICS TS for z/OS 4.1: Recovery and Restart Guide

Page 144
Image 144
IBM SC34-7012-01 manual Logical recovery, Physical recovery