Backout

DFSMStvs logs all updates to recoverable data sets that are modified in a unit of recovery. In this way, if a unit of recovery fails, the changes that were made as part of the unit of recovery can be reversed.

Each instance of DFSMStvs has a private undo log. This log contains the status of the units of recovery in a DFSMStvs instance and the backout records required to back out changes that were made to VSAM recoverable data sets by these units of recovery. Any of the following events can trigger a backout request:

A backout request from the application
An abnormal end of the unit of recovery or the TCB with which it is associated
Detection of an error that DFSMStvs cannot fix, such as an I/O error on one of the data sets

If DFSMStvs detects such an error, it votes against committing the unit of recovery when the application calls prepare.

Should an instance of DFSMStvs fail or abnormally end, all in-flight units of recovery that were using that instance at the time of failure are backed out. The backouts for these units of recovery are not performed at the time of the failure by the failed DFSMStvs instance. The backouts are performed either at the time of the failure or later by a peer recovery instance of DFSMStvs.

Records added to the end of an ESDS are not backed out by DFSMStvs because no such function as ERASE exists for an ESDS. You cannot actually delete a record that has been written over in an ESDS. CICS® has an exit that can modify the record to imply that it has been deleted.