Normal termination
In a normal termination, WebSphere MQ stops all activity in an orderly way. You can stop WebSphere MQ using either quiesce, force, or restart mode. The effects are given in Table 6.
Table 6. Termination using QUIESCE, FORCE, and RESTART Thread type QUIESCE FORCE RESTART Active threads Run to completion Back out Back out New threads Can start Not permitted Not permitted New connections Not permitted Not permitted Not permitted Batch applications are notified if a termination occurs while the application is still connected.
With CICS, a current thread runs only to the end of the unit of recovery. With CICS, stopping a queue manager in quiesce mode stops the CICS adapter, and so if an active task contains more than one unit of recovery, the task does not necessarily run to completion.
If you stop a queue manager in force or restart mode, no new threads are allocated, and work on connected threads is rolled back. Using these modes can create in-doubt units of recovery for threads that are between commit processing phases. They are resolved when WebSphere MQ is reconnected with the controlling CICS, IMS, or RRS subsystem.
When you stop a queue manager, in any mode, the steps are:
- Connections are ended.
- WebSphere MQ ceases to accept commands.
- WebSphere MQ ensures that any outstanding updates to the page sets are completed.
- The DISPLAY USAGE command is issued internally by WebSphere MQ so that the restart RBA is recorded on the z/OS console log.
- The shutdown checkpoint is taken and the BSDS is updated.
Terminations that specify quiesce mode do not affect in-doubt units of recovery. Any unit that is in doubt remains in doubt.