DB2 - Problem description
Problem IC95210 | Status: Closed |
STMM BUFFERPOOL MINIMUM NORMAL PAGE AREA AND CONFIGURED BLOCK PAGE AREA ARE NOT MAINTAINED ON ACTIVATION WITH SCALE BACK | |
product: | |
DB2 FOR LUW / DB2FORLUW / A50 - DB2 | |
Problem description: | |
STMM normally maintains a minimum number of normal bufferpool pages (1000 at this time). This is on top of any configured block page area, which STMM does not tune. On database activation, scaling back may occur when memory is constrained. This process only attempts to maintain an overall size of 1000 for a bufferpool, which may compromise the configuration when there is a block page area configured. The following conditions may result in performance degradation and SQL1218 errors. 1. The bufferpool may initialize with a very small number of normal pages (as low as 32). STMM will try to increase this to the normal minimum size on the next tuning cycle. 2. The block page area may be initialized with less than the configured value. This condition will persist during the current activation as STMM does not tune the block page area. 3. In an STMM/HADR environment, STMM does not actively tune the secondary. If the scaling back occurs on the secondary, the condition of having a small number of normal pages will persist, potentially leading to failures hadr replay processing, resulting in the secondary database marked bad. Since STMM is not tuning, any scaling back on the secondary will also be persistent (configuration changes are written to disk and will be effective on the next activation). Example db2diag.log messages showing scaling back with a block page area configured: 2012-10-30-10.03.31.350919-240 I31939A505 LEVEL: Info PID : 8585406 TID : 1814 PROC : db2sysc 0 INSTANCE: mcornish NODE : 000 DB : TEST APPHDL : 0-42 APPID: *LOCAL.mcornish.121030140331 AUTHID : MCORNISH EDUID : 1814 EDUNAME: db2agent (TEST) 0 FUNCTION: DB2 UDB, buffer pool services, sqlbScaleAutoBPsByFactor, probe:35 MESSAGE : Altering bufferpool "BP1" From: "20000"<automatic> To: "1000"<automatic> 2012-10-30-10.03.31.351426-240 E32445A837 LEVEL: Warning PID : 8585406 TID : 1814 PROC : db2sysc 0 INSTANCE: mcornish NODE : 000 DB : TEST APPHDL : 0-42 APPID: *LOCAL.mcornish.121030140331 AUTHID : MCORNISH EDUID : 1814 EDUNAME: db2agent (TEST) 0 FUNCTION: DB2 UDB, buffer pool services, sqlbSetNumSlotsAndNBPForNewRTBPDef, probe:20 MESSAGE : ADM6069W Buffer pool "BP1" (ID "2") has a NUMBLOCKPAGES value of "10000" which exceeds the maximum supported value for a buffer pool of this size. The NUMBLOCKPAGES value for this buffer pool has been reduced to "980" for its current activation. The value of NUMBLOCKPAGES can be permanently changed by using the ALTER BUFFERPOOL SQL statement. | |
Problem Summary: | |
**************************************************************** * USERS AFFECTED: * * Systems using block-based bufferpools and STMM tuning of * * those bufferpools * **************************************************************** * PROBLEM DESCRIPTION: * * See Error Description * **************************************************************** * RECOMMENDATION: * * Upgrade to DB2 Version 10.5 Fix Pack 3 * **************************************************************** | |
Local Fix: | |
Manually alter the bufferpool configuration to expected values (For HADR, any updates on the primary will flow through to the secondary) | |
available fix packs: | |
DB2 Version 10.5 Fix Pack 3 for Linux, UNIX, and Windows | |
Solution | |
Problem first fixed in DB2 Version 10.5 Fix Pack 3 | |
Workaround | |
see Local Fix | |
Timestamps | |
Date - problem reported : Date - problem closed : Date - last modified : | 27.08.2013 27.02.2014 27.02.2014 |
Problem solved at the following versions (IBM BugInfos) | |
Problem solved according to the fixlist(s) of the following version(s) | |
10.5.0.3 | |
10.5.0.3 |