Over the weekend we upgraded DB2 from version 9 to version 10 NFM, and now in CA Detector I see one of our program's plans showing an extensive abort count.
What could be the cause of this?
Upon examination, using CA Detector with the two datastores provided, the extensive abort counts are revealed.
However what was not shown in CA Detector were SQL errors to correlate with the abort count. This condition can be directly related to 'non-SQL' error conditions such as:
-Application Program Abend
-Application Roll back request
-Application Deadlocked (-911s)
-Application Cancelled By Operator
-Thread Abend Caused By resource Shortage
-In General, any time a unit of recovery is rolled-back
Detector will increment the abort count based on the DB2 Control Block that correlates to IBM field QWACABRT which is part of the IFCID 003 record, including
rollbacks as part of the Abort count. Examine the QWACABRT field using a DB2 monitoring tool such as CA's Sysview Performance Monitor for DB2 for z/OS
(formerly known as Insight for DB2 for z/OS) to get more in-depth information on what could cause CA Detector to reflect the increased abort count associated
with a given plan outside the scope of SQL errors.