DB2 FOR LUW / DB2FORLUW / 910 - DB2

Incorrect results for a query that contains multiple distinct's 
and registry setting for ENHANCED_MULTIPLE_DISTINCT is turned 
on. 
. 
Description: 
If the following conditions are true: 
a) The DB2 instance is configured with DPF (multiple partitions) 
and the registry variable DB2_EXTENDED_OPTIMIZATION is set to 
ENHANCED_MULTIPLE_DISTINCT. 
b) A query is run that contains multiple distinct operations 
(see example) 
c) The query has predicates that need to be evaluated without an 
index.  That is, in order to qualify or disqualify the row 
during predicate evaluation, it has to fetch the row and read 
the columns to compare the predicate rather than reading just 
the index (A sargable predicate). 
Then the query may return incorrect results. 
For example: 
select sum(distinct col2), avg(distinct col3), sum(distinct 
col4) 
from test 
where col5.data = 'abcde' 
group by col1; 
The wrong results is that even rows that should be disqualified 
based on the predicate col5.data = 'abcde', are still included 
in the query incorrectly.  Thus, any sums or averages that are 
performed on the data may have extra data used in those 
aggregate operations.

apr ic69060 
fixed >= v91 fpk10

If all of the local predicates can be resolved by doing 
an index scan as start/stop key predicates, (and not as 
sargable predicates), then this is a possible workaround 
to the problem.

DB2 Version 9.1 Fix Pack 10  for Linux, UNIX and Windows
 DB2 Version 9.1 Fix Pack 11  for Linux, UNIX and Windows
 DB2 Version 9.1 Fix Pack 12  for Linux, UNIX and Windows
 

Incorrect results for a query that contains multiple distinct's 
and registry setting for ENHANCED_MULTIPLE_DISTINCT is turned 
on. 
. 
Description: 
If the following conditions are true: 
a) The DB2 instance is configured with DPF (multiple partitions) 
and the registry variable DB2_EXTENDED_OPTIMIZATION is set to 
ENHANCED_MULTIPLE_DISTINCT. 
b) A query is run that contains multiple distinct operations 
(see example) 
c) The query has predicates that need to be evaluated without an 
index.  That is, in order to qualify or disqualify the row 
during predicate evaluation, it has to fetch the row and read 
the columns to compare the predicate rather than reading just 
the index (A sargable predicate). 
Then the query may return incorrect results. 
For example: 
select sum(distinct col2), avg(distinct col3), sum(distinct 
col4) 
from test 
where col5.data = 'abcde' 
group by col1; 
The wrong results is that even rows that should be disqualified 
based on the predicate col5.data = 'abcde', are still included 
in the query incorrectly.  Thus, any sums or averages that are 
performed on the data may have extra data used in those 
aggregate operations.

If all of the local predicates can be resolved by doing 
an index scan as start/stop key predicates, (and not as 
sargable predicates), then this is a possible workaround 
to the problem.

forerunner  : APAR is sysrouted TO one or more of the following: IC69061 IC69062 IC71085 
follow-up : 

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

07.06.2010
14.06.2011
14.06.2011