Writing the CBRHADUX exit

In general, the routines you code for the auto-delete installation exit should follow these criteria:

Written in Assembler H or High Level Assembler
Handle multiple requests (reentrant)
Reside in SYS1.LINKLIB
Include any valid combination of AMODE and RMODE
Return to the caller using the BSM instruction

The sample auto-delete installation exit has a Sample auto-delete installation exit that prevents objects from being deleted. It also contains code to allow automatic deletion of objects which can be activated with a simple code modification. Once the three lines of code are commented out, the sample auto-delete installation exit reads the VERIFY data set. This is specified at the label, VRFYDSN and needs to be modified based upon your installation. The VERIFY data set should contain objects, along with their collections, that should never be deleted. Try to minimize the number of records in this file because this module is called frequently. From this file the user exit builds an internal table used for each subsequent call. When an object and collection are matched a return code of X'4' and a reason code of X'2', which signifies this object should not be expired.

Attention: Take care to avoid processing overhead because it can affect the time it takes to process the OSMC storage management cycle. For example, when the two lines of bypass code are removed, the sample exit, Sample auto-delete installation exit, reads the names of the objects once per cycle. It maintains the verify table to avoid I/O for each object, and it issues I/O to a data set instead of a teleprocessing link to allow for notification of deleted objects.

If the object name is not found in the verification table, the exit approves deletion. When the object is deleted, the name is written to a sequential data set called the notify data set, HLQ.XXXXXXXX.OBJECT.DELETE.NOTIFY. Fully qualified data set names contain both the object name and the collection name, allowing uniqueness across OSMC and concurrent I/O from the different tasks. A concatenation of these data sets (one for each group) provides input to other applications that need to synchronize their directories with the OAM object directories.