11i - 12 Gather Schema Statistics fails with Ora-20001 errors after 11G database Upgrade (文档 ID 781813.1)

In this Document

APPLIES TO:

Oracle Applications Technology Stack - Version 11.5.10.2 to 12.1.3 [Release 11.5.10 to 12.1]
Oracle EBS Applications Performance - Version 12.1.1 to 12.1.1 [Release 12.1]
Information in this document applies to any platform.
Oracle Server Enterprise Edition - Version: 11.1.0.6 to 11.1.0.8

SYMPTOMS

"Gather Schema Statistics" program reported following errors in request log files:

Error #1: ERROR: While GATHER_TABLE_STATS: 
object_name=GL.JE_BE_LINE_TYPE_MAP***ORA-20001: invalid column name or duplicate columns/column groups/expressions in method_opt*** 
Error #2: ERROR: While GATHER_TABLE_STATS: 
object_name=GL.JE_BE_LOGS***ORA-20001: invalid column name or duplicate columns/column groups/expressions in method_opt***
Error #3: ERROR: While GATHER_TABLE_STATS: 
object_name=GL.JE_BE_VAT_REP_RULES***ORA-20001: invalid column name or duplicate columns/column groups/expressions in method_opt***

Error #4: ERROR: While GATHER_TABLE_STATS:  object_name=FII.FII_FIN_ITEM_HIERARCHIES***ORA-20001: invalid column name or duplicate columns/column groups/expressions in method_opt***

CHANGES

Problem started after database has been upgraded to 11G.

CAUSE

There are two reasons for that error message:

1 ) There are duplicate rows on FND_HISTOGRAM_COLS table for JE_BE_LINE_TYPE_MAP table.
Because of this problem, FND_STATS tries to gather histogram information using wrong command and
it fails with ora-20001 errors.

The following SQL should return one row, not two:

SQL>select column_name, nvl(hsize,254) hsize
from FND_HISTOGRAM_COLS 
where table_name = ‘JE_BE_LINE_TYPE_MAP‘
order by column_name;

COLUMN_NAME HSIZE
------------------------------ ----------
SOURCE 254
SOURCE 254

Since there are two rows in histograms table, FND_STATS creates following command to gather statistics on table ‘JE_BE_LINE_TYPE_MAP‘ :

dbms_Stats.gather_table_stats(OWNNAME => ‘GL‘, TABNAME 
=>‘JE_BE_LINE_TYPE_MAP‘,METHOD_OPT=>‘FOR ALL COLUMNS SIZE 1 FOR COLUMNS SOURCE SIZE 254 FOR 
COLUMNS SOURCE SIZE 254‘);

Above command will work on 9i and 10G databases but it will fail with ora-20001 errors on 11G.

2) Column does not exist on the table but still listed in FND_HISTOGRAMS_COL table.

You can use following SQL to identify. SQL will prompt for table name, use table name from the errors. In above examples you can use FII_FIN_ITEM_HIERARCHIES.

select hc.table_name, hc.column_name
from FND_HISTOGRAM_COLS hc , dba_tab_columns tc
where hc.table_name =‘&TABLE_NAME‘
and hc.table_name= tc.table_name (+)
and hc.column_name = tc.column_name (+)
and tc.column_name is null;

SOLUTION

Find out all duplicates and/or obsolete rows in FND_HISTOGRAM_COLS and delete one of them logged in as the applsys user.
Remember to take backup of the FND_HISTOGRAM_COLS table before deleting any data.

-- identify duplicate rows

select table_name, column_name, count(*)
from FND_HISTOGRAM_COLS
group by table_name, column_name
having count(*) > 1;

-- Use above results on the following SQL to delete duplicates

delete from FND_HISTOGRAM_COLS
where table_name = ‘&TABLE_NAME‘
and  column_name = ‘&COLUMN_NAME‘
and rownum=1;

-- Use following SQL to delete obsoleted rows

delete from FND_HISTOGRAM_COLS
where (table_name, column_name) in 
  (
   select hc.table_name, hc.column_name
   from FND_HISTOGRAM_COLS hc , dba_tab_columns tc
   where hc.table_name  =‘&TABLE_NAME‘
   and hc.table_name= tc.table_name (+)
   and hc.column_name = tc.column_name (+)
   and tc.column_name is null
  );

commit;