DB2 - Problem description
Problem IT02581 | Status: Closed |
TRAPS AND/OR CRASHES IN DPF DURING QUERY ACTIVITY AGAINST A DB THAT WAS PREVIOUSLY CREATED WITH A REDIRECTED RESTORE | |
product: | |
DB2 FOR LUW / DB2FORLUW / A50 - DB2 | |
Problem description: | |
This issue only occurs in a DPF environment. When a database is first created using the "CREATE DB" DDL, DB2 identifies this database with a db_seed. The db_seed is derived as a function of the database creation time and is assumed to be unique for every database in the DPF instance. This assumption is broken when a backed up database is used to create a copy database in the same instance via a redirected restore operation. In this case, it is possible to have two uniquely named databases sharing the same db_seed. Certain database operations that assume uniqueness of the db_seed will be broken. One such operation is the inconsistent maintenance of the participating node list for each database at the catalog node. This node list is involved in ensuring that drop and alter DDL and other invalidating operations (like implicit or explict package rebinds) properly invalidate the package cache and the catalog cache on all relevant partitions. With this issue, one or more partition may not be sent the invalidating operation - leaving stale and invalid entries in either the package cache or the catalog cache. This may potentially lead to outages including traps, sql0901n and sql0902n errors for applications connected to either database that may occur during compilation or in execution of SQL; these errors would have their first point of failure at non-catalog partitions. Another problem that may occur is the successful deactivation of a database may not bring down the database even when user applications are no longer connected. A quick method to determine exposure to this issue is to check the db_seed of two databases. If they are the same, the problem exists. Use the following command to find the db_seed of each database. "db2pd -sqedbcb -db <dbname> | grep db_seed". Similarly, a sample trap that may occur could be the following: <StackTrace> ossDumpStackTraceInternal ossDumpStackTraceV98 OSSTrapFile::dumpEx sqlo_trce sqloEDUCodeTrapHandler sqlktopn sqlriopn sqlriReceiveDss sqlrr_dss_router sqlrr_subagent_router sqleProcessSubRequest sqeAgent::RunEDU sqzEDUObj::EDUDriver sqlzRunEDU sqloEDUEntry </StackTrace> | |
Problem Summary: | |
**************************************************************** * USERS AFFECTED: * * All DPF users * **************************************************************** * PROBLEM DESCRIPTION: * * See Error Description * **************************************************************** * RECOMMENDATION: * * Upgrade to DB2 Cancun Release 10.5.0.4 (also known as Fix * * Pack 4) or higher. * **************************************************************** | |
Local Fix: | |
available fix packs: | |
DB2 Cancun Release 10.5.0.4 (also known as Fix Pack 4) for Linux, UNIX, and Windows | |
Solution | |
Fixed in DB2 Cancun Release 10.5.0.4 (also known as Fix Pack 4) | |
Workaround | |
not known / see Local fix | |
Timestamps | |
Date - problem reported : Date - problem closed : Date - last modified : | 16.06.2014 13.10.2014 13.10.2014 |
Problem solved at the following versions (IBM BugInfos) | |
Problem solved according to the fixlist(s) of the following version(s) | |
10.5.0.4 |