(After a successful redirect restore.) The ROLLFORWARD command
might fail with SQL1042C and you will see "DMS Database
Termination: Fix count != 0" in the db2diag.log.
db2diag.log:
2013-07-19-12.53.50.490554-240 E387755E512 LEVEL: Info
PID : 24638 TID : 140732509710080PROC :
db2sysc 0
INSTANCE: instance1 NODE : 000 DB :
TESTDB
APPHDL : 0-43131 APPID:
*LOCAL.instance1.130719165211
AUTHID : TESTDB2
EDUID : 135 EDUNAME: db2agent (TESTDB) 0
FUNCTION: DB2 UDB, data protection services, sqlufrol,
probe:2120
MESSAGE : ADM1605I DB2 is invoking the backward phase of
database rollforward
recovery.
<<>>
2013-07-19-12.53.50.490797-240 I388268E515 LEVEL:
Warning
PID : 24638 TID : 140732509710080PROC :
db2sysc 0
INSTANCE: instance1 NODE : 000 DB :
TESTDB
APPHDL : 0-43131 APPID:
*LOCAL.instance1.130719165211
AUTHID : TESTDB2
EDUID : 135 EDUNAME: db2agent (TESTDB) 0
FUNCTION: DB2 UDB, recovery manager, sqlpForwardRecovery,
probe:2210
DATA #1 : <preformatted>
Invoking database rollforward backward recovery, nextLso:
6119844005761
<<>>
2013-07-19-12.53.51.034763-240 I388784E509 LEVEL: Info
PID : 24638 TID : 140732509710080PROC :
db2sysc 0
INSTANCE: instance1 NODE : 000 DB :
TESTDB
APPHDL : 0-43131 APPID:
*LOCAL.instance1.130719165211
AUTHID : TESTDB2
EDUID : 135 EDUNAME: db2agent (TESTDB) 0
FUNCTION: DB2 UDB, recovery manager, sqlpForwardRecovery,
probe:2350
MESSAGE : Rollforward is completing. New log chain is:
DATA #1 : unsigned integer, 4 bytes
8
<<>>
2013-07-19-12.54.27.894894-240 I390876E461 LEVEL: Severe
PID : 24638 TID : 140732509710080PROC :
db2sysc 0
INSTANCE: instance1 NODE : 000 DB :
TESTDB
APPHDL : 0-43131 APPID:
*LOCAL.instance1.130719165211
AUTHID : TESTDB2
EDUID : 135 EDUNAME: db2agent (TESTDB) 0
FUNCTION: DB2 UDB, data management, sqldTermDBCB, probe:5900
MESSAGE : DMS Database Termination: Fix count != 0.
This APAR fixes a scenario where rollforward
(1) marks a table as unavailable when we process a NOT LOGGED
INIT log record, and
(2) later we undo a LOAD START log record:
sample DB2 trace:
7684389 | | | sqlptudo entry [eduid 2405 eduname db2agent]
7684390 | | | | sqlpgrlg entry [eduid 2405 eduname db2agent]
LSO = 6119779753664
7684391 | | | | sqlpgrlg exit
7684392 | | | | sqlpWriteLR entry [eduid 2405 eduname
db2agent]
7684393 | | | | | sqlpgResSpace entry [eduid 2405 eduname
db2agent]
7684394 | | | | | sqlpgResSpace exit
7684395 | | | | | sqlpWriteToLog entry [eduid 2405 eduname
db2agent]
7684396 | | | | | sqlpWriteToLog exit
7684397 | | | | sqlpWriteLR data [probe 1000]
7684398 | | | | sqlpWriteLR exit
7684399 | | | | sqlbIsAllowed entry [eduid 2405 eduname
db2agent]
7684400 | | | | sqlbIsAllowed exit
7684401 | | | | sqldFixTCBObj entry [eduid 2405 eduname
db2agent]
7684402 | | | | | sqldLoadTCBObjDesc entry [eduid 2405
eduname db2agent]
7684403 | | | | | sqldLoadTCBObjDesc data [probe 0]
7684404 | | | | | sqldLoadTCBObjDesc exit
7684405 | | | | sqldFixTCBObj exit
corresponding log record:
Record 1493119 Log Page Offset = 07BF = 1983
Log File Offset = 0626E7BF = 103212991
Record LSN = 00000597DD20C7BF
Record LSO (PSO) = 6119779753664
(00000597DD20C7BF)
Record Size = 1618 = 5656
Record Type = 4A = Load Start
Log Header Flags = 0042
Record TID = 000007D913E3
Back Pointer LSO = 0
RecordID = LD02 and Flush Objects
Pool ID = 65530 Object ID = 32774
Table is RP = TRUE
Num pools = 2 Num Objects = 20
LR index = 0 Num LR parts = 1
Update Lsn = FALSE
Object[0] :
ObjID :
poolID = 5
objectID = 536
parentPoolID = 5
parentObjectID = 536
objectType = 0
lifeLSN = 00000597DCA1D0B9
TID = 0:0:0
extentAnchor = 7939552
initEmpPages = 0
objectState = 0x00000027
lastSMP = 0
This APAR might also fix some other similar-looking scenarios
where rollforward fails because of "DMS Database Termination:
Fix count != 0"; but it might not fix all such scenarios
(because we might not be aware of all of them). |
