DB2 FOR LUW / DB2FORLUW / A50 - DB2

After an upgrade to DB2 version 10.5.0.4 or 10.5.0.5, HADR 
environment can experience performance degradation during 
inplace reorg on a standby database.  Symptoms will include 
a large log replay gap on the standby database, indicating 
slow log replay.  Standby system might become congested 
and the log replay might slow down considerably. 
 
To detect this APAR you can run command "db2pd -d <databasename> 
-hadr" on the standby.  You are hitting this APAR if 
STANDBY_REPLAY_LOG_FILE is far behind the STANDBY_LOG_FILE. 
 
You might or might not detect this issue through the health 
monitor : It might report a warning: 
ADM10502W  Health indicator "HADR Operational status" 
("db.hadr_op_status") is in state "Congested" on "database"

**************************************************************** 
* USERS AFFECTED:                                              * 
* DB2 version 10.5.0.4 or 10.5.0.5 and HADR                    * 
**************************************************************** 
* PROBLEM DESCRIPTION:                                         * 
* See Error Description                                        * 
**************************************************************** 
* RECOMMENDATION:                                              * 
* Upgrade to DB2 version 10.5.0.6.                             * 
****************************************************************

Immediate relief can be obtained by turned off the Event 
Recorder trace for this component. 
This only leads to a minimal loss in diagnostic capability in 
the unlikely case of an outage. 
 
This command gives you the EventTypeMask for the event recorder 
:  ( executed on the standby server ) 
 
db2pd -db <db> -dmpevrec comp=SQLD name=OLR query 
 
Component: 4 ( SQLD ) 
Address            CompID Name 
Started  Scope    DBName   EventTypeMask      Level Rsz Wrp 
BufSz ApphdlMask 
0x00007FBD98EC0080 4      OLR                              1 
Database <db>     0xFFFFFFFFFFFFFFFF 65535 0   0      4M 
 
This is set to 0xFFFFFFFFFFFFFFFF  meaning that all events are 
traced. 
Using : 
 
db2pd -db <db>  -dmpevrec comp=SQLD name=OLR mask=0 
 
db_state1 db_state2 db_state3 db_state4 use_count dbname 
dbpath 
--------- --------- --------- --------- --------- -------- 
-------------------- 
000000001 000000000 000000000 000000000 000000032 <DB> 
/home/db2inst1/db2inst1/NODE0000/SQL00001/MEMBER0000/ 
 
SQLD Event recorder OLR mask is set to 0x0000000000000000. 
 
you can turn off EventTypeMask which will make the slow function 
call return immediately without the performance impact is has 
now. 
This would need to be run on the standby .

The problem is first fixed in DB2 version 10.5.0.6.

not known / see Local fix

Date  - problem reported    :
Date  - problem closed      :
Date  - last modified       :

12.01.2015
14.08.2015
21.03.2017