How indoubt units of work are resolved by CICS
One of the functions of the CICS®-WebSphere® MQ adapter is to keep data synchronized between CICS and WebSphere MQ. If a queue manager ends abnormally while connected to CICS, CICS might commit or back out work without WebSphere MQ being aware of it. When the queue manager restarts, that work is indoubt.
WebSphere MQ cannot resolve these indoubt units of work (that is, commit or back out the changes made to WebSphere MQ resources) until the connection to CICS is restarted or reconnected.
- The CICS-WebSphere MQ adapter requests a list of indoubt units of work for this connection ID from WebSphere MQ.
- The adapter receives the list of indoubt units of work, and passes them to CICS for resolution.
- CICS compares entries from this list with entries in its own log. CICS determines from its own list what action it took for each indoubt unit of work.
The resolution of indoubt units of work has no effect on CICS resources. WebSphere MQ considers CICS to be the recovery coordinator, and, when WebSphere MQ restarts, it automatically commits or backs out each unit, depending on whether a log record marked the beginning of the commit. The existence of indoubt objects does not lock CICS resources while WebSphere MQ is being reconnected.
Unresolved units of work
If you are using a WebSphere MQ queue-sharing group for the connection, and you have specified RESYNCMEMBER(NO) on the MQCONN definition for the connection, CICS makes only one attempt to reconnect to the same queue manager. If that attempt fails, CICS connects to any eligible member of the queue-sharing group. A queue manager is eligible for connection to a CICS region if it is currently active on the same LPAR as the CICS region. When CICS connects to a different queue manager, indoubt units of work cannot be resolved, and CICS issues the warning message DFHMQ2064 if any unresolved units of work remain.
- DFHMQ0404E
- DFHMQ0405E
- DFHMQ0406E
- DFHMQ0407E
If any unresolved units remain after restart, resolve them by the methods described in How to resolve CICS units of work manually.
WebSphere MQ supports group units of recovery for CICS, and using WebSphere MQ 7.1, or higher, you can use the new option RESYNCMEMBER(GROUPRESYNC). With the GROUPRESYNC option, CICS connects to any local WebSphere MQ queue manager in the queue-sharing group. The queue manager is chosen by WebSphere MQ and it asks CICS to resolve all indoubt units of work regardless of which queue manager has the indoubt unit of work. Do not change the setting for RESYNCMEMBER when units of work are outstanding in WebSphere MQ as this means that the units of work cannot be resolved. A unit of work held in CICS is identified with a resource manager qualifier. When RESYNCMEMBER(GROUPRESYNC) is used the qualifier is the name of the queue-sharing group, otherwise the qualifier used is the name of the individual queue manager.
- If you specify RESYNCMEMBER(GROUPRESYNC) and the previous connection used an MQCONN definition with RESYNCMEMBER(YES) or RESYNCMEMBER(NO) and there are indoubt units of work outstanding in WebSphere MQ, those indoubt units of work cannot be resolved without reverting to the previous setting of RESYNCMEMBER. CICS issues the warning message DFHMQ2065.
- If you specify RESYNCMEMBER(YES) or RESYNCMEMBER(NO) and the previous connection used an MQCONN definition with RESYNCMEMBER(GROUPRESYNC) and there are indoubt units of work outstanding in WebSphere MQ, those indoubt units of work cannot be resolved without reverting to the previous setting of RESYNCMEMBER. CICS issues the warning message DFHMQ2066.