Home

 

Channels and serialization

 

If a queue manager in a queue-sharing group fails while a message channel agent is dealing with uncommitted messages on one or more shared queues, the channel and the associated channel initiator will end, and shared queue peer recovery will take place for the queue manager.

Because shared queue peer recovery is an asynchronous activity, peer channel recovery might try to simultaneously restart the channel in another part of the queue sharing group before shared queue peer recovery is complete. If this happens, committed messages might be processed ahead of the messages still being recovered. To ensure that messages are not processed out of sequence in this way, message channel agents that process messages on shared queues serialize their access to these queues by issuing the MQCONNX API call.

An attempt to start a channel for which shared queue peer recovery is still in progress might result in a failure. An error message indicating that recovery is in progress is issued, and the channel is put into retry state. Once queue manager peer recovery is complete, the channel can restart at the time of the next retry.

An attempt to RESOLVE, PING, or DELETE a channel can fail for the same reason.

 

Parent topic:

Preparing WebSphere MQ for z/OS for DQM with queue-sharing groups

ic15490_

 

Home