Creating an index on one member(member 0 in this case) in
pureScale failed with error SQL0901N
CREATE UNIQUE INDEX DB2INST1.TAB1_PK ON DB2INST1.TAB1 ( C1 ASC
,C2 ASC ,C3 ASC)
DB21034E The command was processed as an SQL statement because
it was not a
valid Command Line Processor command. During SQL processing it
returned:
SQL0901N The SQL statement or command failed because of a
database system
error. (Reason "Map error to -901".) SQLSTATE=58004
The error in db2diag.log is pointing to a missing log in member
1, even though this log file is archived well before CREATE
INDEX was issued
2018-06-01-00.07.17.955078+600 I15876694E838 LEVEL:
Error
PID : 12345 TID : 140602192684800 PROC :
db2sysc 0
INSTANCE: db2inst1 NODE : 000 DB :
SAMPLE
APPHDL : 0-1459 APPID: xxxx
AUTHID : DB2INST1 HOSTNAME: xxxx
EDUID : 1396 EDUNAME: db2shred.1 (SAMPLE) 0
FUNCTION: DB2 UDB, recovery manager, sqlpshrFindExtentForLfsLsn,
probe:750
MESSAGE : ZRC=0x86100026=-2045771738=SQLPR_MISSING_LOG_EXT
"Used by various operations reading the logs to
indicate a missing log extent."
DATA #1 : String, 8 bytes
SAMPLE
DATA #2 : String, 12 bytes
S0000234.LOG
DATA #3 : Database Partition Number, PD_TYPE_NODE, 2 bytes
0
DATA #4 : db2LogStreamIDType, PD_TYPE_DB2_LOG_STREAM_ID, 2 bytes
1
This log file was automatically purged due to auto_del_rec_obj
database parameter settings.
The reason this problem happens is, there is a mapping scan in
shredder that will find the starting log file for each log
stream to start the shredder scan.
This is an extreme boundary case that is only possible when all
the old log files of the log stream were deleted from the
archive. |
