DB2 FOR LUW / DB2FORLUW / 970 - DB2

Subagent may trap or hang after detecting WLM threshold 
violation while it is in FCM wait.  In the trap case, a subagent 
may trap in function ReceiveBuffer (callstack 1 below).  In the 
hang case, the application will become unresponsive and cannot 
be forced off.  The subagent will remain stuck waiting on a 
waitpost in function WaitSendReady (callstack 2 below) or 
WaitRecvReady (callstack 3 below).  Check to see if the database 
has a reactive threshold that applies to the trap/hung 
application's current activity.  If so, check the event monitor 
table to see if the activity has violated the threshold.  This 
trap/hang is only possible on V9.7 FP5 on DPF instances with 
reactive WLM thresholds enabled. 
 
Callstack 1 - Trap case: 
 
sqkfChannel::ReceiveBuffer 
sqkdBdsBufferTable::getNextBuffer 
sqlkd_rcv_buffer 
sqlkd_rcv_get_next_buffer 
sqlkd_rcv_data 
sqlkdReceiveReply 
sqleInformCoordRequest 
sqleRemoteLobBuffer 
... 
sqlrr_subagent_router 
sqleSubRequestRouter 
sqleProcessSubRequest 
sqeAgent::RunEDU 
 
Callstack 2 - Hang case (WaitSendReady): 
 
sqloWaitEDUWaitPost 
sqkfChannel::WaitSendReady 
sqkfChannel::SendDataBuffer 
sqlkqsnd 
sqlktsnd 
sqlkt_pack_tuple 
sqlktins 
sqlritqb 
sqlriSectInvoke 
sqlrr_dss_router 
sqlrr_subagent_router 
sqleSubRequestRouter 
sqleProcessSubRequest 
sqeAgent::RunEDU 
 
Callstack 3 - Hang case (WaitRecvReady): 
 
sqloWaitEDUWaitPost 
WaitRecvReady 
ReceiveBuffer 
sqlkqrcv 
sqlktrcv 
sqlkt_unpack_tuple 
sqlktftc 
sqlritqa 
sqlriSectInvoke 
sqlrr_dss_router 
sqlrr_subagent_router 
sqleSubRequestRouter 
sqleProcessSubRequest 
sqeAgent::RunEDU

**************************************************************** 
* USERS AFFECTED:                                              * 
* DPF with reactive WLM thresholds                             * 
**************************************************************** 
* PROBLEM DESCRIPTION:                                         * 
* See Error Description                                        * 
**************************************************************** 
* RECOMMENDATION:                                              * 
* Customers on V9.7 FP5 should upgrade to V9.7 FP6 or above.   * 
****************************************************************

The workaround is to kill and restart the instance, and disable 
reactive WLM thresholds.  The permanent fix is to apply V9.7 FP6 
or above.

DB2 Version 9.7 Fix Pack 6 for Linux, UNIX, and Windows
 DB2 Version 9.7 Fix Pack 7 for Linux, UNIX, and Windows
 DB2 Version 9.7 Fix Pack 8 for Linux, UNIX, and Windows
 DB2 Version 9.7 Fix Pack 9 for Linux, UNIX, and Windows
 DB2 Version 9.7 Fix Pack 9a for Linux, UNIX, and Windows
 DB2 Version 9.7 Fix Pack 10 for Linux, UNIX, and Windows
 

This problem only exists in V9.7 FP5.  A fix for this issue is 
delivered to V9.7 FP6.  The hang should no longer be possible 
after fix is applied.

not known / see Local fix

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

27.03.2012
04.06.2012
04.06.2012

9.7.FP5,
9.7.FP6