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 1.
The sequence of TD actions in this example, and the subsequent recovery, is
as follows:
- The unit of work has read items 1 and 2, leaving the read pointer at item 3.
- The unit of work has written item 4, leaving the write pointer ready for the next item to be written.
- CICS abends and is restarted with an emergency restart.
- As 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.
Figure 1. Illustration of recovery of a physically recoverable
TD queue

Making intrapartition TD physically recoverable can be useful in the case of some CICS queues. For example, after a CICS failure, you might choose to restart CICS as quickly as possible, and then look for the cause of the failure. By specifying queues such as CSMT as intrapartition and physically recoverable, the messages produced just before the failure can be recovered and are therefore available to help you diagnose the problem.