Checkpoint

A checkpoint, together with the data written to the recovery log, ensures that all IBM® MQ data, except for nonpersistent messages, is made persistent after a z/TPF system IPL.

The checkpoint process writes all persistent messages in memory queues and some channel data to disk on a regular basis. Except for nonpersistent messages and expired messages that were not saved, a system IPL will rebuild the queues and channels to the same state as before the system IPL by using the IBM MQ checkpoint and the data found on the recovery log.

The recovery log must be big enough to accommodate all logging activity for all z/TPF resource managers (not just the IBM MQ resource manager) between each successful completed checkpoint. The checkpoint interval is 5 seconds. (If your z/TPF system is running under VPARS, the checkpoint interval is 1 second.) If checkpointing is taking too long to complete, a warning message is sent to the COMMS and AUDT functional support consoles. The number of seconds to wait before sending this warning message is specified by entering the ZMQSC DEFINE QMGR command and can be changed by entering the ZMQSC ALTER QMGR command with the CHECKPTIMER parameter specified. By default, the warning message is sent after 5 seconds. For more information about the recovery log, see Recovery log support.

Checkpointing is critical to maintaining IBM MQ on the z/TPF system. The process will be completed even when the system is in input list shutdown. In addition, if the checkpoint process exits prematurely, another checkpoint process can start so that the checkpoint can be completed.