DB2 - Problem description
Problem IT02292 | Status: Closed |
SLOW INDEX INSERT/UPDATE PERFORMANCE IN DB2 ESE FOR DATABASES WHICH HAVE BEEN RESTORED FROM DB2 PURESCALE INSTANCES | |
product: | |
DB2 FOR LUW / DB2FORLUW / A50 - DB2 | |
Problem description: | |
This applies only to databases which have been backed up from a DB2 pureScale instance and restored to a DB2 Enterprise Server Edition instance. Updates/Inserts to very large tables with indexes in these databases may be slow due to searching through all the index space map pages to find a free page in the index object even though there are no free pages available. The reason for this is that the use of Pseudo Used pages in the index object may cause the count of free pages in the index object to fall below zero and thus be treated as a very large number. All the index space map pages are then searched for a free page since it appears as though some free pages exist in the object. No free page will be found so the object will be extended, but the next time a free page is required each space map page will be searched again. This could happen if there were Pseudo Used pages in the index object space map pages after the database is restored to the DB2 Enterprise Server Edition instance. Pseudo Used pages exist naturally in the index object space map pages in DB2 pureScale databases. To determine if this could be the cause of slow performance for inserts/updates into large tables with indexes, the following command may be issued and the resulting binary file can be sent to DB2 Support for analysis: db2pd -db <wsdb> -dmsdbcb If DB2 Support determines that the freePageCnt for the index object of the table(s) is larger than the number of pages in the object then this is likely the cause of the performance issues. | |
Problem Summary: | |
**************************************************************** * USERS AFFECTED: * * User migrating a database from a DB2 pureScale instance to a * * DB2 Enterprise Server * * Edition instance * **************************************************************** * PROBLEM DESCRIPTION: * * See Error Description * **************************************************************** * RECOMMENDATION: * * Upgrade to DB2 version 10.5.0.4. * **************************************************************** | |
Local Fix: | |
Deactivating (DEACTIVATE DATABASE <database>) and re-activating the database (eg REACTIVATE DATABASE <database> or CONNECT TO <database>) will resolve the problem for those tables where all the Pseudo Used pages from the index objects have now been used. To determine if all of the Pseudo Used pages in an index object have been used, issue db2dart command against all the space map pages in the index object. Applying the APAR fix will resolve the problem for all tables. | |
available fix packs: | |
DB2 Cancun Release 10.5.0.4 (also known as Fix Pack 4) for Linux, UNIX, and Windows | |
Solution | |
The problem is first fixed in DB2 version 10.5.0.4. | |
Workaround | |
not known / see Local fix | |
Timestamps | |
Date - problem reported : Date - problem closed : Date - last modified : | 05.06.2014 05.06.2014 05.06.2014 |
Problem solved at the following versions (IBM BugInfos) | |
Problem solved according to the fixlist(s) of the following version(s) | |
10.5.0.4 |