DB2 - Problem description
Problem IC65140 | Status: Closed |
MIGRATION TO V9.7 MAY HANG TRYING TO SCAN CATALOG TABLE SYSIBM.SYSTABLESPACES IF ITS INDEXES ARE MARKED INVALID. | |
product: | |
DB2 FOR LUW / DB2FORLUW / 970 - DB2 | |
Problem description: | |
When migrating to v9.7, the root page of an index is also migrated. If db2 does not find enough space on the root page to migrate it, it will mark the index invalid. These indexes are rebuilt in the later part of the migration. . . Similarly, When there isn't enough space on the root page of the indexes on SYSIBM.SYSTABLESPACES, we mark the indexes invalid. You will see following db2diag.log entry: . . FUNCTION: DB2 UDB, index manager, sqliMigrateRootPage, probe:8796043655424 DATA #1 : <preformatted> Index will be marked to be rebuilt and migration will continue. Root page=5 index IID= 4 TableSpaceID=0 TableId=16 Index_TableSpaceId=0 Index_ObjectId=16 . . In this case tableid=16, tablespaceid=0 refers to table SYSIBM.SYSTABLESPACES. . . Later in the migration where we try to scan the catalog table with the bad index, we get stuck into a loop with the following entries repeating in the db2diag.log: . . FUNCTION: DB2 UDB, data management, sqldLoadTCB, probe:7390^M MESSAGE : ADM5572I One or more indexes on table "TBSPACEID=0.TABLEID=16" are marked invalid and require rebuilding. . | |
Problem Summary: | |
When migrating to v9.7, the root page of an index is also migrated. If db2 does not find enough space on the root page to migrate it, it will mark the index invalid. These indexes are rebuilt in the later part of the migration. . . Similarly, When there isn't enough space on the root page of the indexes on SYSIBM.SYSTABLESPACES, we mark the indexes invalid. You will see following db2diag.log entry: . . FUNCTION: DB2 UDB, index manager, sqliMigrateRootPage, probe:8796043655424 DATA #1 : <preformatted> Index will be marked to be rebuilt and migration will continue. Root page=5 index IID= 4 TableSpaceID=0 TableId=16 Index_TableSpaceId=0 Index_ObjectId=16 . . In this case tableid=16, tablespaceid=0 refers to table SYSIBM.SYSTABLESPACES. . . Later in the migration where we try to scan the catalog table with the bad index, we get stuck into a loop with the following entries repeating in the db2diag.log: . . FUNCTION: DB2 UDB, data management, sqldLoadTCB, probe:7390^M MESSAGE : ADM5572I One or more indexes on table "TBSPACEID=0.TABLEID=16" are marked invalid and require rebuilding. | |
Local Fix: | |
REORG or recreate indexes on SYSIBM.SYSTABLESPACES before the migration. This will create enough space on the root page to allow the migration or the root page. Hence, the index will not be marked invalid. | |
available fix packs: | |
DB2 Version 9.7 Fix Pack 2 for Linux, UNIX, and Windows | |
Solution | |
Defect => wsdbu00604952 Fixed in => V9.7 + FP2 Module => ENGN_SQRL | |
Workaround | |
REORG or recreate indexes on SYSIBM.SYSTABLESPACES before the migration. This will create enough space on the root page to allow the migration or the root page. Hence, the index will not be marked invalid. | |
Timestamps | |
Date - problem reported : Date - problem closed : Date - last modified : | 14.12.2009 10.06.2010 10.06.2010 |
Problem solved at the following versions (IBM BugInfos) | |
9.7. | |
Problem solved according to the fixlist(s) of the following version(s) | |
9.7.0.2 |