Oracle 11.2.0.3 handles the problem of continuous growth of sysaux table space without downtime

appearance

SYSAUX Table space has been growing , A month of rapid growth 10G, Table spaces need to be maintained regularly

reason

oracle bug cause (Doc ID 1055547.1)

MMON performs the purge of the optimizer stats history automatically. However it has an internal limit of 5 minutes to perform this job. If the operation takes more than 5 minutes, then it is aborted and stats not purged.
No trace or alert message is reported.

Processing mode - No patches

Disable statistics

Because when collecting statistical information, you need to insert data into the related table below , Statistics cannot be collected during cleanup .

wri$_optstat_tab_history

wri$_optstat_ind_history

wri$_optstat_histhead_history

wri$_optstat_histgrm_history

Sign in ebs Put two （ One GL、 One ALL） Plan request Statistical data mode Cancel it （ Cancel the previous screenshot backup parameters ）.

Delete historical data

All of the following sql in sys as dba perform

First, calculate the total earliest date of the existing data .

SELECT TRUNC(SYSDATE) - TO_DATE(to_char(MIN(savtime), 'YYYY-MM-DD'), 'yyyy-mm-dd')
  FROM sys.WRI$_OPTSTAT_HISTHEAD_HISTORY;

The assumption is 190 God , Then log in through remote desktop sqlplus （ The following sql It may take several days , You need to run in the background or remote desktop ）

set serveroutput on
BEGIN
DBMS_OUTPUT.ENABLE(buffer_size => null);
for i in reverse 10..190
loop
dbms_output.put_line(to_char(sysdate,'YYYY-MM-DD HH24:MI:SS') || '-Begin delete Day ' || i );
dbms_stats.purge_stats(sysdate-i);
COMMIT;
dbms_output.put_line(to_char(sysdate,'YYYY-MM-DD HH24:MI:SS') || '-Complete delete Day ' || i );
end loop;
end;
/

Above this sql It may take several days , After execution, use shink Shrink the table .

Shrink the table

-- First back up the index data 
select dbms_metadata.get_ddl('INDEX','I_WRI$_OPTSTAT_HH_OBJ_ICOL_ST','SYS') from dual;
select dbms_metadata.get_ddl('INDEX','I_WRI$_OPTSTAT_HH_ST','SYS') from dual;

-- Drop indexes 
drop index I_WRI$_OPTSTAT_HH_OBJ_ICOL_ST;
drop index I_WRI$_OPTSTAT_HH_ST;

-- Enable row migration 
alter table WRI$_OPTSTAT_HISTHEAD_HISTORY enable row movement; 

-- To contract （ It's useless here move To save space ）
alter table WRI$_OPTSTAT_HISTHEAD_HISTORY shrink space cascade;

-- Index reconstruction after shrinking 
CREATE UNIQUE INDEX "SYS"."I_WRI$_OPTSTAT_HH_OBJ_ICOL_ST" ON "SYS"."WRI$_OPTSTAT_HISTHEAD_HISTORY" ("OBJ#", "INTCOL#", SYS_EXTRACT_UTC("SAVTIME"), "COLNAME") 
  PCTFREE 10 INITRANS 2 MAXTRANS 255 COMPUTE STATISTICS 
  STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
  PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1
  BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
  TABLESPACE "SYSAUX" ;


CREATE INDEX "SYS"."I_WRI$_OPTSTAT_HH_ST" ON "SYS"."WRI$_OPTSTAT_HISTHEAD_HISTORY" (SYS_EXTRACT_UTC("SAVTIME")) 
  PCTFREE 10 INITRANS 2 MAXTRANS 255 COMPUTE STATISTICS 
  STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
  PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1
  BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
  TABLESPACE "SYSAUX" ;

Collect statistics

EXEC dbms_stats.gather_table_stats(ownname => ‘SYS’,tabname => ‘WRI$_OPTSTAT_HISTHEAD_HISTORY’,cascade => TRUE);

See the effect

SELECT *
  FROM (SELECT owner, segment_name, segment_type, SUM(bytes) / 1024 / 1024 / 1024 GB
          FROM dba_segments
         WHERE tablespace_name = 'SYSAUX'
         GROUP BY owner, segment_name, segment_type
         ORDER BY 4 DESC)
 WHERE rownum < 10;

SELECT occupant_name "Item",
       space_usage_kbytes / 1048576 "Space Used (GB)",
       schema_name "Schema",
       move_procedure "Move Procedure"
  FROM v$sysaux_occupants
 ORDER BY space_usage_kbytes DESC;

Resume statistical mode information collection

Submit it immediately ALL Statistics of patterns , Then set a plan request ,ALL Pattern , Five nights a week 10 Click to collect data , Parameter reference backup screenshot .
Because the weekly data size of the general ledger model changes little , To save power , There is no need to set a schedule to request running every night .
Not EBS The library needs to use the corresponding api To enable and disable statistics collection .