Unit of recovery disposition on z/OS

Certain transactional applications can use a GROUP, rather than a QMGR, unit of recovery disposition when connected to a queue manager in a queue sharing group (QSG) by specifying the QSG name when they connect instead of the queue manager name. This allows transaction recovery to be more flexible and robust by removing the requirement to reconnect to the same queue manager in the QSG.

Transactions started by applications that have connected using the queue sharing group name also have a GROUP unit of recovery disposition.

When a transactional application connects with a GROUP unit of recovery disposition it is logically connected to the queue sharing group and does not have an affinity to any specific queue manager. Any 2-phase commit transactions that it has started that have completed phase-1 of the commit process, that is, they are in doubt, can be inquired and resolved, when connected to any queue manager within the QSG. In a recovery scenario this means that the transaction coordinator does not have to reconnect to the same queue manager, which may be unavailable at that time.

Applications that connect with a QMGR unit of recovery disposition have a direct affinity to the queue manager to which they are connected. In a recovery scenario the transaction coordinator must reconnect to the same queue manager to resolve any in-doubt transactions, irrespective of whether the queue manager belongs to a queue sharing group.

When applications specify a queue sharing group name, and thus connect to a queue manager in a QSG with a GROUP unit of recovery disposition, the queue sharing group is logically a separate resource manager. This means that in-doubt transactions are only visible to an application if it reconnects with the same unit of recovery disposition. In-doubt transactions with a QMGR unit of recovery disposition are not visible to applications that have connected with a GROUP unit of recovery disposition and vice versa.