Protecting the resource

When the resource manager processes the UR, it needs to prepare for changes to its resource, which includes two steps:

Log the unchanged data
Log the potential changed data.

If all resource managers vote YES to indicate that they can make the changes, RRS instructs each resource manager to make the changes; only then does the resource manager make the changes. If any resource manager votes NO to indicate that it cannot make the changes, RRS instructs all resource managers to not make the changes; the resource managers backout the changes.

The services used to protect the resource are:

Callable service	Description
Retrieve_UR_Data	Retrieve UR data, including the UR identifier (URID).
Set_Persistent_Interest_Data	Provide persistent interest data.
Retrieve_UR_Interest_Data	Retrieve persistent and nonpersistent interest data.
Post_Deferred_UR_Exit	Give to RRS the response from a resource manager exit routine that previously replied with a defer return code. A later response might be needed in a distributed environment.

For information on the services, see Callable resource recovery services.

The exit routines used to protect the resource are:

Exit Routine	Event for Invoking the Routine
STATE_CHECK	An application program called the Application_Commit_UR service.
PREPARE	RRS is ready for the resource managers to prepare for resource recovery.
ONLY_AGENT	Only one resource manager expressed only one interest in the UR.
DISTRIBUTED_SYNCPOINT	When a resource manager has taken the DSRM role for an interest in a UR and the UR becomes in-doubt, this exit resolves the in-doubt condition
COMMIT	The resource manager is to make the changes permanent.
BACKOUT	The resource manager is to not make the changes, thus, backing them out.
END_UR	The UR processing is finished.
EXIT_FAILED	An RRS exit routine in the resource manager failed.
COMPLETION	The exit routine can do processing needed before RRS returns control to the application program.
SUBODINATE_FAILED	Either RRS or any resource manager on a subordinate system failed, the subordinate system itself terminated, or the context associated with the subordinate UR abnormally terminated.

For information on the exit routines, see Resource recovery exit routines.