PI33243: ABND04E RC00E72018 AT DSNXSRME OFFSET1F68 UNDER SPECIFIC SCENARIOS WHEN QUERY INCLUDE DISTINCT AND GROUP BY

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• ABND04E RC00E72018 at DSNXSRME OFFSET1F68 might be issued under
  specific scenarios if DISTINCT KEY and GROUP BY key are
  pre-ordered as input to sort.
  
  The query may looks like this:
  
  select Count(distinct a), b from t1
  group by b order by 1 des;
  

Local fix

• N/A
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All DB2 11 for z/OS users of queries that    *
  *                 involve distinct and group by keys which are *
  *                 pre-ordered as input to sort.                *
  ****************************************************************
  * PROBLEM DESCRIPTION: An ABEND04E RC00E72018 can occur at     *
  *                      DSNXSRME+1F68 (OFFSET1F68) under        *
  *                      specific scenarios when a query         *
  *                      involves distinct and group by keys     *
  *                      which are pre-ordered as input to sort. *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  An ABEND04E RC00E72018 at DSNXSRME OFFSET1F68 can occur under
  specific scenarios when a query involves distinct and group by
  keys which are pre-ordered as input to sort.  This will usually
  involve a lot of data which causes sort to create multiple runs
  (workfiles) during its input phase.
  
  Below is an example query that could cause this problem.
  
  Step 1. Create Table T1 with index IX2.
  
    CREATE TABLE T1 (A CHAR(1), B VARCHAR(20));
  
    CREATE INDEX IX2 ON T1(B, A);
  
  
  Step 2. Insert many rows of data.
    Insert lots of rows (need sort to create multiple workfiles).
  
  
  Step 3. Perform the following query with DISTINCT and GROUP BY.
  
    SELECT A, COUNT(DISTINCT B), MAX(B) FROM T1 GROUP BY A;
  
  
  Given the above scenario, sort will have the 'total size' of the
  sort key (A + B) and the size of the sort key to compare (A),
  'comparison size', when doing its calculation.  In most
  scenarios, the size of the sort key and comparison key are the
  same.  However, when finding the key in the workfile, sort needs
  to use the 'total size' of the key to get to its offset instead
  of the 'comparison size'.
  

Problem conclusion

• DB2 has been modified to address the issue about the sort key
  size for comparison and for finding it (by using the correct
  offset) in the workfile.
  
  Additional Keywords: SQLSORT SQLGROUPBY SQLDISTINCT
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

  UI25404

Modules/Macros

•    DSNXSRME
  

Fix information

• Fixed component name

  DB2 OS/390 & Z/

• Fixed component ID

  5740XYR00

Applicable component levels

• RB10 PSY UI25404

     UP15/03/11 P F503

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.