MQRC 2162: Queue manager stopping

What it means

Once a queue manager is stopping, every MQI call on the connection fails with MQCC_FAILED and reason 2162. If the call was an MQGET waiting with MQGMO_WAIT, the wait is cancelled. This happens in an immediate or preemptive shutdown, including one an administrator issues to speed up a controlled shutdown that is taking too long.

The application doesn't need to back out work coordinated by the queue manager, because the queue manager backs it out. If an external transaction manager coordinates the unit of work, the application should issue that manager's back-out call. For client applications, IBM warns that the failing call may have completed even though it reported failure, the same caution as for 2009.

On z/OS, timing decides which code an application sees. Depending on how far the shutdown has got, a call may return 2009 (MQRC_CONNECTION_BROKEN) instead of 2162. Code that handles one should handle the other the same way.

Common causes

• An administrator stopped the queue manager immediately, for example with endmqm -i or endmqm -p on distributed MQ.
• On z/OS, an operator issued STOP QMGR MODE(FORCE).
• A failover moved the queue manager to another node and the old instance is ending.
• The queue manager is ending because of a serious error, and the error log or SYSLOG shows why.

How to find the problem

1. Check the queue manager error log, or SYSLOG and the MSTR job log on z/OS, for the stop command or the failure that started the shutdown.
2. Run dspmq on distributed MQ to see the current state.
3. In a high availability setup, check which node now runs the queue manager and whether the failover completed.
4. For clients, compare the time of the 2162 with the server-side shutdown messages to confirm the link.

How to fix it

• End the application cleanly, or close down its MQ work and wait to reconnect.
• If an external transaction manager coordinates the work, issue its back-out call.
• Build reconnect logic, or use automatic client reconnect, so the application resumes when the queue manager returns.
• Make the program safe to repeat its last call on a client connection, because that call may have completed.
• Handle 2162 and 2009 the same way in the code that deals with shutdown. Handle 2161 apart from them, because after 2161 the program can still finish its current unit of work.

Technical reviewer: Albert Joseph, WND Software.

Last reviewed 2026-10-18 by the WND Software team. Written in our own words. For the full reference, check IBM’s MQ documentation for your version.