Troubleshooting
Problem
Several memory messages in the db2diag.log which are aparently being generated by a process called kuddb2
Symptom
Memory management block allocation failure in db2diag.log from kuddb2 process'
Cause
The kuddb2 process' are the agents of IBM Tivoli Monitoring (ITM) product which could be installed as part of some other IBM products.
And the messages are caused by ITM tuning issue
Environment
DB2 product installed along with IBM Tivoli Monitoring for Databases
Diagnosing The Problem
There will be several messages similar to the following in the db2diag.log which will show those are coming from kuddb2 process'
2011-07-17-13.42.35.020506-240 I2406200A1804 LEVEL: Warning
PID : 5767352 TID : 7199 PROC : kuddb2
INSTANCE: <inst-name> NODE : 000
EDUID : 7199
FUNCTION: DB2 UDB, SQO Memory Management, sqlogmblkEx, probe:1000
MESSAGE : ZRC=0x820F0004=-2112946172=SQLO_MEM_SIZE "Mem Mgt invalid size"
DIA8563C An invalid memory size was requested.
DATA #1 : String, 43 bytes
Memory management block allocation failure.
DATA #2 : Codepath, 8 bytes 2
DATA #3 : Memory pool handle pointer, PD_TYPE_MEM_POOL_HANDLE_PTR, 8 bytes 0x0000000000000000
DATA #4 : Requested size, PD_TYPE_MEM_REQUESTED_SIZE, 8 bytes 0
DATA #5 : Adjusted block size, PD_TYPE_MEM_ADJUSTED_SIZE, 8 bytes 0
DATA #6 : Options for requested block, PD_TYPE_GET_MEM_OPTIONS, 4 bytes 0x00000000
DATA #7 : Pointer to address that will be set by new allocation,
PD_TYPE_PTR_TO_ADDRESS_OUT, 8 bytes
0x0000000111058f20
DATA #8 : File name, PD_TYPE_OSS_MEM_FILE_NAME, 9 bytes sqlmapi.C
DATA #9 : Line of code, PD_TYPE_OSS_MEM_LINE_NUM, 8 bytes 2918
DATA #10: Resource binding pointer, PD_TYPE_RESOURCE_BINDING_PTR, 8 bytes
0x0000000000000000
CALLSTCK:
[0] 0x09000000014F2D70 pdLog + 0xA8
[1] 0x090000000153FBB8 pdLog@glue21E + 0x254
[2] 0x09000000014FD6AC sqlogmblkEx + 0x7F4
[3] 0x09000000012E60F4
@84@sqlm_send_est_buf_size_db2ra__FUiPUiPvsT1P15sqlmStreamFlagsP5sqlca + 0x4FC
[4] 0x0900000001A4BB4C db2GetSnapshotSize + 0xB48
[5] 0x0000000100020514
GetDatabaseSnapshot__17CandleDb2SnapShotFPiiT1N22PPcT2 + 0x358
[6] 0x00000001000A9344 TakeSample__21kud00_kudtabspc_agentFv + 0x518
[7] 0x090000000389C654 DriveDataCollection__5ctiraFv + 0x268
[8] 0x09000000038A9014 checkForExpiredRequests__12TableManagerFl + 0x400
[9] 0x09000000038AAB34 timeout__12TableManagerFP15CTRA_Timerspec_ + 0x38
Resolving The Problem
These memory messages from kuddb2 process' are not DB2 engine problem.
These shows the ITM agents are experiencing memory issues for some reasons.
If users want to get rid of these messages immediately and don't care any effects on the ITM monitoring then those process' could be killed at UNIX level with "kill -9 <process-id>" for the kuddb2 process'
One possibility of the root cause of the problem could be,
In some versions of ITM products sometimes the agent don't take the correct
information from the parameter CTIRA_HOSTNAME and can lose connection.
To fix that,
1) First stop the agent.
Then, set the parameter CTIRA_HOSTNAME with the actual server hostname in the ud.ini config file.
(The ud.ini file could be found as <itm-install-dir>/config/ud.ini )
The hostname should be less than or equal to 32 characters.
CTIRA_HOSTNAME=HOSTNAME
( hostname could be obtained by running "hostname" at the command prompt)
2) Then Reconfigure the agent :
./itmcmd config -A pc ( pc = product code, for DB2 it's "ud" )
3) When the agent finalize the reconfiguration verify that the changes
in CTIRA_HOSTNAME on the ud.ini file are still set.
Also, check the ITM user has unlimited data in ulimit output at UNIX level.
For any further help on these steps the IBM Tivoli Monitoring support could be contacted.
To start and stop the ITM agent for DB2 :
DB2_DIR/bin/itmcmd agent -o instance stop ud
DB2_DIR/bin/itmcmd agent -o instance start ud
Related Information
Was this topic helpful?
Document Information
Modified date:
16 June 2018
UID
swg21507079