DB2 - Problem description
Problem IC82650 | Status: Closed |
LOAD RESTART MIGHT REMOVE WRONG DUPLICATE ROW DURING DELETE PHASE, OR BUILD CORRUPTED XML INDEX | |
product: | |
DB2 FOR LUW / DB2FORLUW / 980 - DB2 | |
Problem description: | |
A LOAD RESTART on a table with unique index, during the load delete phase (when we delete duplicate rows) might incorrectly delete old rows instead of new rows. The result is incorrect data in the table after the LOAD. A LOAD RESTART on a table with XML column might build XML index with incorrect RID (i.e. causing index corruption). Subsequent operations that access the table would fail, with messages in db2diag.log referring to index error and possibly sqlischd, for example 2012-03-03-10.56.01.379104-300 I482734651E1581 LEVEL: Severe PID : 11483 TID : 46916050086208 PROC : db2sysc 1 INSTANCE: regress1 NODE : 001 DB : ABSINTHE APPHDL : 0-3701 APPID: 9.26.96.75.33218.120303163346 AUTHID : REGRESS1 HOSTNAME: vmx3650svt32e EDUID : 4717 EDUNAME: db2agntp (ABSINTHE) 1 FUNCTION: DB2 UDB, index manager, sqlischd, probe:973 RETCODE : ZRC=0x8709002C=-2029453268=SQLI_NOKEY "Key not found within node" DIA8541C The index key could not be found, the value was "". DATA #1 : String, 11 bytes blatch_mode DATA #2 : signed integer, 4 bytes 3 DATA #3 : String, 11 bytes lkey_option DATA #4 : signed integer, 4 bytes 8 DATA #5 : String, 11 bytes lsn_changed DATA #6 : signed integer, 4 bytes 0 DATA #7 : String, 10 bytes node_level DATA #8 : signed integer, 4 bytes 1 DATA #9 : String, 14 bytes recursive_call DATA #10: signed integer, 4 bytes 0 DATA #11: String, 5 bytes retry DATA #12: signed integer, 4 bytes 0 DATA #13: String, 9 bytes saved_lsn DATA #14: SQLP_LSN8, PD_TYPE_SQLP_LSN8, 8 bytes 00000000027932F8 DATA #15: String, 8 bytes slotinfo DATA #16: Hexdump, 40 bytes 0x00002AAB7E7FBC30 : 0000 0000 0000 0000 1A00 0000 0000 0000 ................ 0x00002AAB7E7FBC40 : 1400 0000 0000 0000 1A00 0000 0000 0000 ................ 0x00002AAB7E7FBC50 : 0000 0000 0000 0000 ........ DATA #17: String, 7 bytes pagenum DATA #18: unsigned integer, 4 bytes 89 DATA #19: String, 9 bytes currlevel DATA #20: signed integer, 2 bytes 1 2012-03-03-10.56.01.385992-300 E482736233E829 LEVEL: Error PID : 11483 TID : 46916050086208 PROC : db2sysc 1 INSTANCE: regress1 NODE : 001 DB : ABSINTHE APPHDL : 0-3701 APPID: 9.26.96.75.33218.120303163346 AUTHID : REGRESS1 HOSTNAME: vmx3650svt32e EDUID : 4717 EDUNAME: db2agntp (ABSINTHE) 1 FUNCTION: DB2 UDB, index manager, sqliDumpPageFromAllMembers, probe:20 MESSAGE : ADM14005E The following error occurred: "IndexError". First Occurrence Data Capture (FODC) has been invoked in the following mode: "Automatic". Diagnostic information has been recorded in the directory named "/home/regress1/sqllib/db2dump/FODC_IndexError_2012-03-03-10.56. 01.38 3373_11483_4717_001/". | |
Problem Summary: | |
**************************************************************** * USERS AFFECTED: * * All users using LOAD RESTART * **************************************************************** * PROBLEM DESCRIPTION: * * See Error Description * **************************************************************** * RECOMMENDATION: * * Upgrade to DB2 Version 9.8 Fix Pack 5. * **************************************************************** | |
Local Fix: | |
Avoid doing LOAD RESTART; do LOAD TERMINATE and then LOAD INSERT or LOAD REPLACE instead | |
Solution | |
Problem was first fixed in DB2 Version 9.8 Fix Pack 5. | |
Workaround | |
not known / see Local fix | |
Timestamps | |
Date - problem reported : Date - problem closed : Date - last modified : | 13.04.2012 13.06.2012 13.06.2012 |
Problem solved at the following versions (IBM BugInfos) | |
9.8.FP5 | |
Problem solved according to the fixlist(s) of the following version(s) | |
9.8.0.5 |