DB2 - Problem description
Problem IT30681 | Status: Closed |
APPLICATION MEMORY CONTINUALLY GROWS IN DPF ENVIRONMENTS AFTER UPGRADING TO RECENT FIX PACK LEVELS | |
product: | |
DB2 FOR LUW / DB2FORLUW / A10 - DB2 | |
Problem description: | |
A problem similar to a memory leak has been introduced in the following levels of code : 10.1 Fix Pack 6 (APAR IT18654) 10.5 Fix Pack 9 (APAR IT18655) 11.1 Mod 1 Fix Pack 1 (APAR IT18656) 11.5 GA Only DPF environments with long-running transactions are vulnerable. The memory buildup occurs when an SQL statement is repeatedly invoked without committing. On remote nodes only (not the coordinator), memory allocations associated with the SQL section are freshly allocated on each invocation instead of being reused. Upon commit, however, the memory is released (cached for reuse or freed as part of normal cache fluctuations). There are several factors that point to this problem being the source of increased memory usage: 1. The problem occurs after upgrading to the above-mentioned levels 2. The memory usage is in the "Application Memory" consumer : db2pd -dbptnmem (ensure to run on remote partitions, eg. -member ) Name Mem Used (KB) HWM Used (KB) Cached (KB) ======================================================== APPL-TEST 160000 160000 157632 db2 "select * from table(mon_get_memory_set('APPLICATION','TEST',-2))" will show increasing values for several columns, including MEMORY_SET_SIZE 3. The specific memory pool that is increasing in memory usage is the "Application Shared" (APPL_SHARED) heap : db2 "select * from table(mon_get_memory_pool('APPLICATION','TEST',-2)) where MEMORY_POOL_TYPE='APPL_SHARED'" In db2pd this is the "appshr" pool in db2pd -db -mempools output) 4. specific memory blocks increasing are similar to the following : db2pd -db -memblocks top appl 20 Top memory consumers in Appl memory set: PoolID PoolName TotalSize(Bytes) %Bytes TotalCount %Count LOC File File hash code 20 appshrh 555048 52.37 39 14.39 5063 sqlra_sql_workspace.C 1095929873 20 appshrh 86112 8.12 39 14.39 1193 sqlriSectInit.C 1127556507 20 appshrh 20592 1.94 39 14.39 7550 sqlri_dss.C 930532034 20 appshrh 19968 1.88 39 14.39 1556 sqlriSectInit.C 1127556507 20 appshrh 4992 0.47 39 14.39 2490 sqlra_sql_workspace.C 1095929873 20 appshrh 4680 0.44 39 14.39 1116 sqlriSectInit.C 1127556507 5. Applications are seen with long-running transactions, eg. the following monitoring statement shows a very high number of activity requests within a unit of work : db2 "select uow_id,act_rqsts_total from table(mon_get_unit_of_work(null,null)) order by act_rqsts_total desc" | |
Problem Summary: | |
**************************************************************** * USERS AFFECTED: * * DPF configurations with long-running transactions * **************************************************************** * PROBLEM DESCRIPTION: * * See Error Description * **************************************************************** * RECOMMENDATION: * * Upgrade to next Release, see version 11.1 APAR IT31700 * **************************************************************** | |
Local Fix: | |
Solution | |
Workaround | |
**************************************************************** * USERS AFFECTED: * * DPF configurations with long-running transactions * **************************************************************** * PROBLEM DESCRIPTION: * * See Error Description * **************************************************************** * RECOMMENDATION: * * Upgrade to next Release, see version 11.1 APAR IT31700 * **************************************************************** | |
Timestamps | |
Date - problem reported : Date - problem closed : Date - last modified : | 23.10.2019 17.03.2021 17.03.2021 |
Problem solved at the following versions (IBM BugInfos) | |
Problem solved according to the fixlist(s) of the following version(s) |