SYSAUX表空间满的解决方法

转载自：https://www.cnblogs.com/yiyuf/p/4284448.html

step1.  确认到底是哪个段占用了sysaux空间：

select segment_name,sum(bytes)/1024/1024 from dba_segments where tablespace_name='SYSAUX' group by segment_name order by 2 desc

SEGMENT_NAME 　　　　　　　　　　　　　　SUM(BYTES)/1024/1024
WRH$_ACTIVE_SESSION_HISTORY 　　　　7360.375
WRH$_ACTIVE_SESSION_HISTORY_PK　　 1400.257

从上面看就是ASH（v$active_session_history）历史记录没有清理造成

step2. 检查有多少条无效记录：

SQL> SELECT COUNT(1) Orphaned_ASH_Rows
 FROM wrh$_active_session_history a
 WHERE NOT EXISTS
 (SELECT 1
 FROM wrm$_snapshot
 WHERE snap_id = a.snap_id
 AND dbid = a.dbid
 AND instance_number = a.instance_number
 );

ORPHANED_ASH_ROWS
-----------------
23392228 

step3.清楚多余记录，由于rows过多，清理速度有点慢。

SQL>DELETE FROM wrh$_active_session_history a
WHERE NOT EXISTS (SELECT 1
FROM wrm$_snapshot
WHERE snap_id = a.snap_id
AND dbid = a.dbid
AND instance_number = a.instance_number);

23392228 rows deleted.

SQL>commit;

期间需要注意归档日志的产生量，避免造成磁盘空间不足等麻烦。同时可以适当的多添加几组redolog来加速delete操作。

step4. shrink tablespace

alter table WRH$_ACTIVE_SESSION_HISTORY shrink space;

step5. re-check

column OCCUPANT_NAME format a15
SELECT occupant_name,
occupant_desc,
space_usage_kbytes
FROM v$sysaux_occupants
WHERE occupant_name LIKE '%AWR%';

SELECT PARTITION_NAME FROM DBA_IND_PARTITIONS WHERE INDEX_NAME='WRH$_ACTIVE_SESSION_HISTORY_PK';

PARTITION_NAME
------------------------------
WRH$_EVENT_HISTO_MXDB_MXSN
WRH$_ACTIVE_2645903699_6157

step6 rebuild index

SQL> alter index sys.WRH$_ACTIVE_SESSION_HISTORY_PK rebuild partition WRH$_ACTIVE_2645903699_12505;

Index altered.

SQL> alter index sys.WRH$_ACTIVE_SESSION_HISTORY_PK rebuild partition WRH$_ACTIVE_2645903699_6157;

Index altered.