DB2 Version 9.7 for Linux, UNIX, and Windows

Enhanced resilience to errors and traps reduces outages

Improvements in the Version 9.7 infrastructure enhance its resilience to certain categories of critical errors and traps. For example, a DB2® instance can remain operational when certain critical data errors are encountered when reading from disk or in-memory data pages, as well as when unexpected traps occur.

In Version 9.7, there are an increased number of cases in which a DB2 instance continues to safely operate even when critical errors and traps occur, and all the while providing diagnostic messages that help in tracking these events. The following additional messages can be generated in the administration notification log and should be monitored by database administrators:

ADM6006E
ADM6007C
ADM14011C
ADM14012C
ADM14013C

In addition, the following SQLCODEs can be received by applications:

SQL1655C
SQL1656C

Remaining operational during peak business hours is the primary objective of every business. The enhanced resiliency of a DB2 instance, in the face of certain categories of critical errors and traps, helps to achieve this objective. If a database instance needs to be recycled after the occurrence of a sustained critical error or trap, these short outages can now be scheduled after peak business hours.

Starting in Fix Pack 3, trap resilience, is automatically applied to the load utility. In the event of a trap during a load operation, the trap will be sustained, if possible, and the table which was being loaded remains in a load pending state, just as it would if any other type of load failure occurred. Although it is possible to recover from the trap by issuing a LOAD RESTART, TERMINATE, or REPLACE command, the instance should still be recycled as soon as possible because the trapped thread and its descendents are left suspended. In previous releases, traps could not be sustained if the load utility was running and the load engine dispatchable units (EDUs) would crash the instance.