Excessive AWR Growth From Partitioned Objects Such as SYS.WRH$_EVENT_HISTOGRAM Causing Sysaux to Grow

AWR数据增长较快，导致sysaux表空间使用较高

SQL> select f.tablespace_name,
  2         a.total,
  3         f.free,
  4         round((f.free / a.total) * 100) "% Free"
  5    from (select tablespace_name, sum(bytes / (1024 * 1024)) total
  6            from dba_data_files
  7           group by tablespace_name) a,
  8         (select tablespace_name, round(sum(bytes / (1024 * 1024))) free
  9            from dba_free_space
 10           group by tablespace_name) f
 11   WHERE a.tablespace_name = f.tablespace_name(+)
 12   order by "% Free"
 13  /

TABLESPACE_NAME                     TOTAL       FREE     % Free
------------------------------ ---------- ---------- ----------
UNDOTBS1                             5185          6          0
SYSAUX                               8660        413          5
USERS                                   5          4         80
SYSTEM                               4640       3798         82
UNDOTBS2                              365        335         92
X_DATA                               1000        990         99
X_INDEX                              1000        986         99

查看是哪些数据库对象占用了sysaux表空间

SQL> select *
  2    from (select segment_name,
  3                 PARTITION_NAME,
  4                 segment_type,
  5                 bytes / 1024 / 1024
  6            from dba_segments
  7           where tablespace_name = 'SYSAUX'
  8           order by 4 desc )
  9   where rownum <= 20;

SEGMENT_NAME                   PARTITION_NAME                 SEGMENT_TYPE       BYTES/1024/1024
------------------------------ ------------------------------ ------------------ ---------------
WRH$_EVENT_HISTOGRAM_PK        WRH$_EVENT__1402125233_0       INDEX PARTITION                736
WRH$_EVENT_HISTOGRAM           WRH$_EVENT__1402125233_0       TABLE PARTITION                658
WRH$_LATCH                     WRH$_LATCH_1402125233_0        TABLE PARTITION                588
WRH$_SQLSTAT                   WRH$_SQLSTA_1402125233_0       TABLE PARTITION                560
WRH$_SYSSTAT_PK                WRH$_SYSSTA_1402125233_0       INDEX PARTITION                504
WRH$_LATCH_PK                  WRH$_LATCH_1402125233_0        INDEX PARTITION                440
WRH$_SYSSTAT                   WRH$_SYSSTA_1402125233_0       TABLE PARTITION                368
WRH$_PARAMETER_PK              WRH$_PARAME_1402125233_0       INDEX PARTITION                336
WRH$_PARAMETER                 WRH$_PARAME_1402125233_0       TABLE PARTITION                280
WRH$_LATCH_MISSES_SUMMARY_PK   WRH$_LATCH__1402125233_0       INDEX PARTITION                280
WRH$_SEG_STAT                  WRH$_SEG_ST_1402125233_0       TABLE PARTITION                256
WRH$_SYSTEM_EVENT              WRH$_SYSTEM_1402125233_0       TABLE PARTITION                232
WRH$_LATCH_MISSES_SUMMARY      WRH$_LATCH__1402125233_0       TABLE PARTITION                224
WRH$_SYSTEM_EVENT_PK           WRH$_SYSTEM_1402125233_0       INDEX PARTITION                216
WRH$_SQLSTAT_PK                WRH$_SQLSTA_1402125233_0       INDEX PARTITION                160
WRH$_SERVICE_STAT_PK           WRH$_SERVIC_1402125233_0       INDEX PARTITION                152
WRH$_DLM_MISC                  WRH$_DLM_MI_1402125233_0       TABLE PARTITION                120
WRH$_ACTIVE_SESSION_HISTORY    WRH$_ACTIVE_1402125233_0       TABLE PARTITION                120
WRH$_SQLSTAT_INDEX             WRH$_SQLSTA_1402125233_0       INDEX PARTITION                112
WRH$_SEG_STAT_PK               WRH$_SEG_ST_1402125233_0       INDEX PARTITION                104

查看awr快照保留期限。设置的期限是8天，但是真正保留了258天

SQL> select sysdate - a.sample_time ash,
  2         sysdate - s.begin_interval_time snap,
  3         c.RETENTION
  4    from sys.wrm$_wr_control c,
  5         (select db.dbid, min(w.sample_time) sample_time
  6            from sys.v_$database db, sys.Wrh$_active_session_history w
  7           where w.dbid = db.dbid
  8           group by db.dbid) a,
  9         (select db.dbid, min(r.begin_interval_time) begin_interval_time
 10            from sys.v_$database db, sys.wrm$_snapshot r
 11           where r.dbid = db.dbid
 12           group by db.dbid) s
 13   where a.dbid = s.dbid
 14     and c.dbid = a.dbid;

ASH                                      SNAP                                     RETENTION
---------------------------------------- ---------------------------------------- -------------------------------------
+000000258 08:55:03.824                  +000000008 08:55:30.192                  +00008 00:00:00.0

也可以通过dba_hist_snapshot和WRH$_EVENT_HISTOGRAM进行比较

SQL> select min(snap_id),MAX(snap_id) from dba_hist_snapshot;

MIN(SNAP_ID) MAX(SNAP_ID)
------------ ------------
        5489         6350

SQL> select min(snap_id),MAX(snap_id) from WRH$_EVENT_HISTOGRAM;

MIN(SNAP_ID) MAX(SNAP_ID)
------------ ------------
         168         6350

SQL>

通过MOS查看，这是数据库的一个bug：

　　Document 14084247.8 Bug 14084247 - Failed AWR purge can lead to continued SYSAUX space use

因为purge操作失败，split分区表没有成功。所以导致分区大小不断增加，进而影响性能（purge job超时，导致AWR过期数据没有被清空）。

但是，即使是安装了补丁后，也不是立即生效，可能需要等一段时间才能生效。

安装了补丁后，还要按照以下步骤执行purge操作：

1) drop expired partitions (DDL)

2) purge by callback (anything can happen here, but mostly no-ops or regular DELETE operations)

3) purge regular non-partitioned tables (i.e, DELETE)

4) split partitions (DDL)