Hi there,

About a week back, we upgraded our 9.5ga server with fixpak 2a. From then on, the db2diag.log file is bloating as if it is gulping mcdonald burgers. The file is getting as big as 7 GB in a single day and all this with the diaglevel 3. I am even scarred to think about what will happen at diaglevel 4!!

The following error messages repeat themselves in the log file and is the reason for the bloating of the file. But before that, the OSLEVEL and DB2LEVEL details are below.

It is an AIX 5.3 Machine
Known Service Packs
-------------------
5300-06-02-0727
5300-06-01-0722
5300-06-00-0000
5300-05-CSP-0000
5300-05-06-0000
5300-05-05-0000
5300-05-04-0000
5300-05-03-0000
5300-05-02-0000
5300-05-01-0000
5300-04-CSP-0000
5300-04-03-0000
5300-04-02-0000
5300-04-01-0000
5300-03-CSP-0000

DB2LEVEL:
~~~~~~
DB21085I Instance "db295i3" uses "64" bits and DB2 code release "SQL09052"
with level identifier "03030107".
Informational tokens are "DB2 v9.5.0.2", "s080911", "U820796", and Fix Pack
"2a".
Product is installed at "/opt/IBM/db2/V9.5ga".

The error messages in the db2diag.log file is as follows:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


2008-11-06-03.04.36.915991+300 I4183A840 LEVEL: Error
PID : 1810578 TID : 5303 PROC : db2sysc 0
INSTANCE: db295i3 NODE : 000
EDUID : 5303 EDUNAME: db2pfchr (PAYLESS) 0
FUNCTION: DB2 UDB, oper system services, sqloReadVLow, probe:100
MESSAGE : ZRC=0x8300006D=-2097151891
DATA #1 : Codepath, 8 bytes
3
DATA #2 : String, 17 bytes
preadv() failed !
DATA #3 : signed integer, 8 bytes
-1
DATA #4 : unsigned integer, 4 bytes
109
DATA #5 : File handle, PD_TYPE_SQO_FILE_HDL, 8 bytes
0x0700000196BFE030 : 0000 00C4 0000 0308 ........
DATA #6 : signed integer, 4 bytes
15
DATA #7 : signed integer, 4 bytes
0
DATA #8 : File Offset, 8 bytes
2599813120
DATA #9 : signed integer, 4 bytes
13
DATA #10: signed integer, 4 bytes
8192
DATA #11: signed integer, 8 bytes
0

2008-11-06-03.04.36.938689+300 I5024A531 LEVEL: Error
PID : 1810578 TID : 5303 PROC : db2sysc 0
INSTANCE: db295i3 NODE : 000
EDUID : 5303 EDUNAME: db2pfchr (PAYLESS) 0
FUNCTION: DB2 UDB, buffer pool services, sqlbReadAndReleaseBuffers, probe:50
MESSAGE : ZRC=0x8300006D=-2097151891
DATA #1 : <preformatted>
Tablespace 3 (LLMDATA) flags=2102 HighWaterMark=5045920 poolPageNumber=4294967295 ConOffset=317360 startppg=3173536 lowestindex=0 highextindex=15 num-pages 16

2008-11-06-03.04.36.938960+300 I5556A337 LEVEL: Error
PID : 1810578 TID : 5303 PROC : db2sysc 0
INSTANCE: db295i3 NODE : 000
EDUID : 5303 EDUNAME: db2pfchr (PAYLESS) 0
FUNCTION: DB2 UDB, buffer pool services, sqlbErrorHandler, probe:0
MESSAGE : ZRC=0x8300006D=-2097151891

2008-11-06-03.04.36.939158+300 I5894A435 LEVEL: Error
PID : 1810578 TID : 5303 PROC : db2sysc 0
INSTANCE: db295i3 NODE : 000
EDUID : 5303 EDUNAME: db2pfchr (PAYLESS) 0
FUNCTION: DB2 UDB, buffer pool services, sqlbErrorHandler, probe:0
DATA #1 : String, 106 bytes
Obj={pool:3;obj:21;type:0} State=x27 Parent={3;21}, EM=128, PP0=144
Prefetcher Error, in sqlbProcessRange

2008-11-06-03.04.36.939326+300 I6330A860 LEVEL: Error
PID : 1810578 TID : 5303 PROC : db2sysc 0
INSTANCE: db295i3 NODE : 000
EDUID : 5303 EDUNAME: db2pfchr (PAYLESS) 0
FUNCTION: DB2 UDB, buffer pool services, sqlbErrorHandler, probe:0
DATA #1 : Object descriptor, PD_TYPE_SQLB_OBJECT_DESC, 72 bytes
Obj: {pool:3;obj:21;type:0} Parent={3;21}
lifeLSN: 0201511701A8
tid: 0 0 0
extentAnchor: 128
initEmpPages: 0
poolPage0: 144
poolflags: 2102
objectState: 27
lastSMP: 0
pageSize: 8192
extentSize: 16
bufferPoolID: 2
partialHash: 1376259
bufferPool: 0x07000001c0849aa0

The only difference between the repetitions is the tablespace.

Your help is highly appreciated.

Thanks,
Ravee

You can archive db2diag.log file by "db2diag -A". With this the file will be archived with current date & new db2diag.log file will be created whenever new log is generated.

I am not sure about this error but i think you should check the Bufferpool & tablespace size.

We do not need the log files and hence we are currently deleting them. However this is an issue since the space we have for log files is very little.

Two more things I forgot to mention in my initial post.

1. All the bufferpools in the database are enabled for self tuning.

2. All the DMS tablespaces have autoresize yes.

Guys,

This is resolved.

The issue was, some where, some process was looking for the tools database "toolsdb" and since it was not present, it was logging in the db2diag and hence the file size increase. However the messages in the log file do not relate to this condition and hence misleading. You can either upgrade to the latest fixpak in which this is fixed, or just create a tools db.

Thanks
Ravee

You can also use Linux command logrotate to automatically delete old log entries. I use this command on db2diag.log file.