Quiesce IBM MQ on IBM i

Stop all queue managers. If necessary force all queue managers to stop, tidy up shared memory and end all jobs in the QMQM subsystem.

About this task

The orderly shutdown of IBM® MQ is called quiescing. You need to quiesce IBM MQ to upgrade to a newer version.

To quiesce one or more queue managers:

Procedure

Prepare to quiesce queue managers:

Sign on to a new interactive IBM i session, ensuring that you are not accessing any IBM MQ objects.
Ensure that you have the following authorities:
1. *ALLOBJ authority, or object management authority for the QMQM library.
2. Sufficient authority to use the ENDSBS command.
Warn all users that you are going to stop IBM MQ.

Quiesce all queue managers:

Run the ENDMQM command:
```
ENDMQM MQMNAME(*ALL) OPTION(*CNTRLD) ENDCCTJOB(*YES) RCDMQMIMG(*YES)
TIMEOUT( 15 )
```
Where 15 is a timeout value in seconds.
If the ENDMQM command has not completed within a reasonable period (at least 10 minutes), use the WRKMQM command. This command identifies the queue managers that are still ending. Then force each one in turn to stop by issuing:
```
ENDMQM MQMNAME( QMGRNAME ) OPTION(*IMMED)
```
Where QMGRNAME is the name of the queue manager.
Complete the tidying up of shared memory by issuing the command:
```
ENDMQM MQMNAME(*ALL) OPTION(*IMMED) ENDCCTJOB(*YES) RCDMQMIMG(*NO)
TIMEOUT( 15 )
```

If the commands in the previous step do not complete, end the subsystem immediately:

Run the following command:
```
ENDSBS SBS(QMQM) OPTION(*IMMED)
```

If the command in the previous step also does not complete, use the operating system command ENDJOB to end all jobs in the subsystem QMQM:

Note: Do not use ENDJOBABN unless you intend to perform an IPL on the machine before starting IBM MQ. Ending IBM MQ jobs using ENDJOBABN can lead to damaged semaphores, which in turn can prevent your queue manager from starting.

If a QMGR must be shut down manually, end the jobs (ENDJOB) in the following order. Wait a few minutes for AMQA* or AMQZ* jobs to tidy up.
1. RUNMQLSR - TCP listener (multi-threaded)
2. AMQCLMAA - TCP listener (single-threaded)
3. AMQRMPPA - Channel process pooling job
4. RUNMQCHI - channel initiator
5. AMQCRSTA - receiving MCA jobs
6. RUNMQCHL - sending MCA jobs
7. AMQCRS6B - LU62 receiver channel
8. AMQPCSEA - command server
9. RUNMQTRM - Application trigger monitor
10. RUNMQDLQ - Dead letter queue handler
11. AMQFCXBA - IBM Integration Bus Worker Job
12. AMQFQPUB - Queued Publish/Subscribe Daemon
13. RUNMQBRK - IBM Integration Bus Control Job
14. AMQZMUC0 ('0' is a zero) - Utility Manager
15. AMQZMUF0 ('0' is a zero) - Utility Manager
16. AMQZMUR0 ('0' is a zero) - Utility Manager
17. AMQZMGR0 ('0' is a zero) - Process Controller
18. AMQRRMFA - cluster repository manager
19. AMQZDMAA - deferred message manager
20. AMQALMPX - Log Manager
21. AMQZFUMA - object authority manager
22. AMQZLSA0 ('0' is a zero) - LQM agents
23. AMQZLAA0 ('0' is a zero) - LQM agents
24. AMQZXMA0 ('0' is a zero) - Execution Controller

Run the following command:


ENDMQM MQMNAME( QMGRNAME ) OPTION(*IMMED)

Run the following command:


ENDMQM MQMNAME(*ALL) OPTION(*CNTRLD) ENDCCTJOB(*YES) RCDMQMIMG(*NO)
TIMEOUT( 05 )

Where 05 is a timeout value in seconds.

Manually clean up shared memory.
Run the following command:
```
EDTF '/QIBM/UserData/mqm/qmgrs'
```
then:
1. Take option 5 for &SYSTEM and check that the following directories are empty: isem, esem, msem, ssem, and shmem.
2. Take option 5 for QMGRNAME and check that the following directories are empty:- isem, esem, msem, ssem, and shmem.
3. Take option 5 for &ipcc in the QMGRNAME directory and check that the following directories are empty:- isem, esem, msem, ssem, and shmem.
4. Take option 5 for &qmpersist in the QMGRNAME directory and check that the following directories are empty:- isem, esem, msem, ssem, and shmem.
5. Take option 5 for &app and check that the following directories are empty: isem, esem, msem, ssem, and shmem.