|
Problem IC69772
|
Status: Closed |
POTENTIAL CORRUPTION WHEN REPLAYING LOG RECORDS THAT INSERT KEYS INTO AN INDEX AND TRIGGER PAGE SPLITS |
| product: |
DB2 FOR LUW / DB2FORLUW / 970 - DB2 |
| Problem description: |
As a consequence of replaying log records with the intent of
inserting index keys to trigger page splits, some pages may
become corrupted. Failure to set specific bits in the page
header causes the index manager to behave incorrectly, leading
to inaccurate results and table-to-index mismatches. This
happens during a replay of crash recovery log records due to a
failed LOAD command that uses INDEXING MODE INCREMENTAL (either
explicitly or through the INDEXING MODE AUTOSELECT option).
However, if a new page is loaded into the bufferpool during this
timeframe then the wrong page may be updated, resulting in
inaccurate output. In addition, high availability disaster
recovery (HADR) databases may be affected when replaying log
records on standby systems.
The corruption occurs 53 bytes from the start of the page and
may require setting the 0x40 bit. It could also involve setting
or clearing the 0x10 bit or the 0x20
bits.
These errors occur during crash recovery or database
roll-forward operations.
The INSPECT command, when run with INDEXDATA option, can be used
to detect if your system is affected. |
| Problem Summary: |
****************************************************************
* USERS AFFECTED: *
* All *
****************************************************************
* PROBLEM DESCRIPTION: *
* Replaying log records that trigger an index page split may *
* cause unexpected errors. *
****************************************************************
* RECOMMENDATION: *
* Upgrade to the latest fix pack. *
**************************************************************** |
| Local Fix: |
If an index page gets corrupted, rebuild the index. If a table
page gets corrupted, restore a backup image.
As an alternative, upgrade to v9.7 fixpack 4 or higher and
restart the database (i.e. db2 restart db DBNAME where DBNAME
is the database name). |
| available fix packs: |
DB2 Version 9.7 Fix Pack 3 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 3a for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 4 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 5 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 6 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 7 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 8 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 9 for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 9a for Linux, UNIX, and Windows
DB2 Version 9.7 Fix Pack 10 for Linux, UNIX, and Windows
|
| Solution |
Problem is first fixed in DB2 UDB Version 9.7 fix pack 3. |
| Workaround |
not known / see Local fix |
| BUG-Tracking |
forerunner : APAR is sysrouted TO one or more of the following: IC69812
follow-up : |
| Timestamps |
Date - problem reported :
Date - problem closed :
Date - last modified :
| 08.07.2010
23.09.2010
05.07.2011 |
| Problem solved at the following versions (IBM BugInfos) |
9.7.FP3
|
| Problem solved according to the fixlist(s) of the following version(s) |
| 9.7.0.3
|
 |
| 9.7.0.3
|
|
