DB2 - Problembeschreibung
Problem IC69885 | Status: Geschlossen |
On Windows platforms, LOAD with CPU_PARALLELISM more than 1 may leak message queue memory | |
Produkt: | |
DB2 FOR LUW / DB2FORLUW / 950 - DB2 | |
Problembeschreibung: | |
On Windows, LOAD operations with cpu_parallelism greater than 1 (implicitly or explicitly) can cause message queue memory resource leakage. Even though each LOAD only leaks a small amount of memory (typically less than 1 KB), this Queue Memory is not growable. For heavy LOAD users running thousands of LOADs cumulatively, the Queue Memory can eventually be exhausted, and after that DB2 can experience errors, panics, and hangs. Indicator in db2diag.log is, messages that show: MESSAGE : ZRC=0x870F0041=-2029060031=SQLO_QUE_NOT_SENT "Message Not Sent" | |
Problem-Zusammenfassung: | |
**************************************************************** * USERS AFFECTED: * * Users running large amount of load operations in * * cumulativefashion under windows platform * **************************************************************** * PROBLEM DESCRIPTION: * * In windows platform, users who need to run large amount * * ofDB2 load operations concurrently can experiencea memory * * leak situation from message queue memory resource.This is * * applicable when cpu_parallelism is greater than 1. * **************************************************************** * RECOMMENDATION: * * Customer should upgrade to DB2 V9.5 fix pak 7 to * * avoidexperiencing this defect.Also, the guidelines mentioned * * under LOCAL FIX of thisAPAR could be followed to avoid * * this issue. * **************************************************************** | |
Local-Fix: | |
Reduce the load's cpu_parallelism to 1 (which will prevent leakage from load); Increase the queue memory with DB2 registry DB2NTMEMSIZE=QUE:<size> (which will delay exhaustion of queue memory) | |
verfügbare FixPacks: | |
DB2 Version 9.5 Fix Pack 7 for Linux, UNIX, and Windows | |
Lösung | |
First Fixed in DB2 Version V9.5 fix pak 7 With this fix in place customer will be able to run large number of load operations simultaneously under windows platform. | |
Workaround | |
keiner bekannt / siehe Local-Fix | |
Bug-Verfolgung | |
Vorgänger : APAR is sysrouted TO one or more of the following: IC71348 IC71349 Nachfolger : | |
Weitere Daten | |
Datum - Problem gemeldet : Datum - Problem geschlossen : Datum - der letzten Änderung: | 13.07.2010 25.10.2010 25.10.2010 |
Problem behoben ab folgender Versionen (IBM BugInfos) | |
9.5. | |
Problem behoben lt. FixList in der Version | |
9.1.0.7 | |
9.5.0.7 |