|
Problem IC83130
|
Status: Closed |
INDEX CORRUPTION IN PURESCALE MODE |
| product: |
DB2 FOR LUW / DB2FORLUW / 980 - DB2 |
| Problem description: |
In pureScale, different members have their own transaction log
records with their own log sequence numbers (LSNs).
Periodically we must synchronize the log sequence numbers across
all of the members in the pureScale cluster. There is a case
where we failed to do a log synchronize of the log sequence
numbers during create index and that can lead to a corruption of
the index when dropping and then recreating an index that is not
the last index on the table. For the corruption to occur, there
must be updates to the table made from another member
immediately before the drop index occurs and that member must
crash and require member crash recovery. |
| Problem Summary: |
****************************************************************
* USERS AFFECTED: *
* pureScale environment *
****************************************************************
* PROBLEM DESCRIPTION: *
* See Error Description *
****************************************************************
* RECOMMENDATION: *
* Upgrade to DB2 version 9.8.0.5. *
**************************************************************** |
| Local Fix: |
Run REORG INDEXES ALL with the ALL NO ACCESS option to recreate
the affect index object. |
| Solution |
The problem is first fixed in DB2 version 9.8.0.5. |
| Workaround |
not known / see Local fix |
| Timestamps |
Date - problem reported :
Date - problem closed :
Date - last modified :
| 04.05.2012
14.06.2012
14.06.2012 |
| Problem solved at the following versions (IBM BugInfos) |
9.8.0.5
|
| Problem solved according to the fixlist(s) of the following version(s) |
| 9.8.0.5
|
 |
