Emergency Restart

An emergency restart runs automatically when CICS is restarted after an abnormal end - a region abend, a cancel, or a power failure. Its job is to repair the damage: back out work that was in flight and complete work that was committed.

What emergency restart does

• The recovery manager reads the system log backwards and forwards to find every unit of work that was active when CICS died.
• Units of work that had not reached syncpoint are backed out, so their partial changes disappear.
• Units of work that had committed are completed (forward recovery), so no committed change is lost.
• CICS also resynchronizes with connected systems - remote CICS regions, DB2 and MQ - to settle in-doubt units of work.

How it is triggered

• It is automatic: restart the region after an abnormal end with START=AUTO and CICS detects the abnormal end and runs emergency restart.
• You can also force it explicitly:-
  //CICSSTEP EXEC PGM=DFHSIP,
  // PARM='SI,START=AUTO,APPLID=CICSPROD'
• During the restart, operators may see WTOR messages asking for decisions - for example, whether to proceed when a journal is unavailable. Reply carefully.

After emergency restart

• Check for in-doubt or failed units of work with the operator command CEMT INQUIRE UOW.
• Any UOW shown as FAILED or INDOUBT needs attention - resynchronize it or force it after verifying the data.
• Verify that critical files are open and enabled, and that application teams confirm data integrity before users return.