Group units of recovery (GROUPUR) for CICS
The IBM MQ GROUPUR for CICSĀ® provides peer recovery for in-doubt units of work in a queue sharing group (QSG). One IBM MQ queue manager can resolve in-doubt units of work on behalf of another queue manager in the queue sharing group. This means that if CICS reconnects through group attach to a different queue manager in the QSG, it can resolve indoubt transactions from a previous IBM MQ connection.
If a CICS region is working with a queue manager, and the queue manager ends abnormally, then any indoubt transactions are recovered. This eliminates the need for the CICS region to wait for the queue manager that it was working with to restart, and then resolve any in doubt units of work. This means that you need at least two queue managers on the LPAR, so that CICS can connect to another queue manager in the event of an abnormal termination of the first queue manager.
The new RESYNCMEMBER(GROUPRESYNC) setting on the CICS MQCONN definition:
- Uses the IBM MQ group attach function and peer recovery.
- Requires a queue manager with the GROUPUR attribute enabled.
- Still supports the existing CICS MQCONN RESYNCMEMBER settings (YES and NO):
- Uses the existing CICS group attach function and no peer recovery.
- Change RESYNCMEMBER settings takes effect next time CICS connects to IBM MQ.