以前写了一篇ORACLE临时表空间总结的文章, 里面介绍了几个查看临时表空间使用情况的脚本,其中一个脚本如下所示:
SELECT TU.TABLESPACE_NAME AS "TABLESPACE_NAME",
TT.TOTAL - TU.USED AS "FREE(G)",
TT.TOTAL AS "TOTAL(G)",
ROUND(NVL(TU.USED, 0) / TT.TOTAL * 100, 3) AS "USED(%)",
ROUND(NVL(TT.TOTAL - TU.USED, 0) * 100 / TT.TOTAL, 3) AS "FREE(%)"
FROM (SELECT TABLESPACE_NAME,
SUM(BYTES_USED) / 1024 / 1024 / 1024 USED
FROM GV_$TEMP_SPACE_HEADER
GROUP BY TABLESPACE_NAME) TU ,
(SELECT TABLESPACE_NAME,
SUM(BYTES) / 1024 / 1024 / 1024 AS TOTAL
FROM DBA_TEMP_FILES
GROUP BY TABLESPACE_NAME) TT
WHERE TU.TABLESPACE_NAME = TT.TABLESPACE_NAME;
其实这个查看表空间的脚本是不准确的,有问题的(当然前面博客里面所提到的脚本到现在也没有改,以后也不打算修改了,就这样放着吧)。你可以对比下面几个脚本来看看。
如果临时表空间是dictionary managed temporary tablespace,可以使用下面SQL:
SELECT (S.TOT_USED_BLOCKS/F.TOTAL_BLOCKS)*100 AS "PERCENT USED"
FROM
(SELECT SUM(USED_BLOCKS) TOT_USED_BLOCKS
FROM V$SORT_SEGMENT
WHERE TABLESPACE_NAME='TEMPSCM2'
) S,
(SELECT SUM(BLOCKS) TOTAL_BLOCKS
FROM DBA_DATA_FILES
WHERE TABLESPACE_NAME='TEMPSCM2'
) F;
如果临时表空间是Locally Manageed Temporary Tablespace,使用下面SQL:
SQL> SELECT T.TABLESPACE_NAME,
( U.TOT_USED_BLOCKS / T.TOTAL_BLOCKS ) * 100 AS "PERCENT USED"
FROM (SELECT TABLESPACE_NAME,
SUM(USED_BLOCKS) TOT_USED_BLOCKS
FROM V$SORT_SEGMENT
WHERE TABLESPACE_NAME = &TABLESPACE_NAME
GROUP BY TABLESPACE_NAME) U,
(SELECT TABLESPACE_NAME,
SUM(BLOCKS) TOTAL_BLOCKS
FROM DBA_TEMP_FILES
WHERE TABLESPACE_NAME = &TABLESPACE_NAME
GROUP BY TABLESPACE_NAME) T;
当然你也可以使用下面SQL来查看临时表空间的使用情况, 如下所示:
SELECT D.tablespace_name,
SPACE "SUM_SPACE(M)",
blocks "SUM_BLOCKS",
used_space "USED_SPACE(M)",
Round(Nvl(used_space, 0) / SPACE * 100, 2) "USED_RATE(%)",
SPACE - used_space "FREE_SPACE(M)"
FROM (SELECT tablespace_name,
Round(SUM(bytes) / ( 1024 * 1024 ), 2) SPACE,
SUM(blocks) BLOCKS
FROM dba_temp_files
GROUP BY tablespace_name) D,
(SELECT tablespace,
Round(SUM(blocks * 8192) / ( 1024 * 1024 ), 2) USED_SPACE
FROM v$sort_usage
GROUP BY tablespace) F
WHERE D.tablespace_name = F.tablespace(+)
AND D.tablespace_name='TEMPSCM2'
那么为什么GV_$TEMP_SPACE_HEADER统计的数据不准确呢? 这个是因为GV_$TEMP_SPACE_HEADER取数据不准确,官方解释为:
The views v$sort_usage or v$tempseg_usage ( and v$sort_segment) give the correct information regarding the allocation of sort segments.
We should always query these views to find out the actual temp usage. The view v$temp_space_header shows that these many blocks were touched in each temp file at some point when temp usage was at its highest,
in essence, it shows the number of initialized blocks for each tempfile, not the actual allocated blocks.
The views v$sort_usage/v$tempseg_usage show the actual sort extents allocated for each transaction from these initialized blocks. Also, v$temp_space_header is persistent across restarts. V$sort_segment and v$sort_usage are not.
第二段我翻译如下:
视图v$temp_space_header显示的是每一个temp文件在某一个时刻使用过的最大大小,从本质上说,它显示的是每一个tempfile的初始化大小,而不是实际分配的块大小。
所以说从视图v$temp_space_header获取的数据其实并不是实际使用的大小,它是不准确的。那么肯定有人会问,脚本里面不是访问的GV_$TEMP_SPACE_HEADER视图吗? 跟这个视图v$temp_space_header有关系吗? 答案是有关系,他们的数据来源是一致的,也就是说来自相同的内部表,如下所示:
一般来说,在GV$和V$之后, Oracle会建立GV_$和V_$视图, 随后为这些视图建立了公用同义词。GV_$TEMP_SPACE_HEADER是一个视图,如下所示
SQL> SELECT OWNER, OBJECT_NAME, OBJECT_TYPE FROM DBA_OBJECTS WHERE OBJECT_NAME ='GV_$TEMP_SPACE_HEADER';
OWNER OBJECT_NAME OBJECT_TYPE
-------------------- ------------------------------- -------------------
SYS GV_$TEMP_SPACE_HEADER VIEW
GV_$TEMP_SPACE_HEADER视图的定义如下所示:
SELECT DBMS_METADATA.GET_DDL('VIEW', 'GV_$TEMP_SPACE_HEADER', 'SYS') FROM DUAL;
SELECT * FROM DBA_VIEWS WHERE VIEW_NAME='GV_$TEMP_SPACE_HEADER';
SELECT "INST_ID",
"TABLESPACE_NAME",
"FILE_ID",
"BYTES_USED",
"BLOCKS_USED",
"BYTES_FREE",
"BLOCKS_FREE",
"RELATIVE_FNO"
ROM gv$temp_space_header
而gv$temp_space_header视图的定义如下(当然如果查询DBA_OBJECTS会发现它是一个同义词,指向GV_$TEMP_SPACE_HEADER,这个后面介绍原因)
SQL> select view_definition from v$fixed_view_definition
2 where view_name='GV$TEMP_SPACE_HEADER';
VIEW_DEFINITION
--------------------------------------------------------------------------------
select /*+ ordered use_nl(hc) */ hc.inst_id, ts.name, hc.ktfthctfno, (hc.ktfthcs
z - hc.ktfthcfree)*ts.blocksize, (hc.ktfthcsz - hc.ktfthcfree), hc.ktfthcfree*ts
.blocksize, hc.ktfthcfree, hc.ktfthcfno from ts$ ts, x$ktfthc hc where ts.conten
ts$ = 1 and ts.bitmapped <> 0 and ts.online$ = 1 and ts.ts# = hc.ktfthctsn and h
c.ktfthccval = 0
SQL>
v$temp_space_header它也是一个视图(查询DBA_OBJECTS发现其是一个同义词,这个后面介绍),你会发现v$temp_space_header其实是从视图GV$TEMP_SPACE_HEADER过滤数据,如下所示:
SELECT OWNER, OBJECT_NAME, OBJECT_TYPE FROM DBA_OBJECTS WHERE OBJECT_NAME =upper('v$temp_space_header');
QL> select view_definition from v$fixed_view_definition
2 where view_name=upper('v$temp_space_header');
IEW_DEFINITION
-------------------------------------------------------------------------------
elect TABLESPACE_NAME , FILE_ID , BYTES_USED , BLOCKS_USED , BYTES_FREE , BLOC
S_FREE , RELATIVE_FNO from GV$TEMP_SPACE_HEADER where inst_id = USERENV('Instan
e')
你在$ORACLE_HOME/rdbms/admin下的catspace.sql中,就会发现这样的SQL,这就解释了为什么gv$temp_space_header是视图,又是同义词的原因。
create or replace view gv_$temp_space_header as select * from gv$temp_space_header;
create or replace public synonym gv$temp_space_header
for gv_$temp_space_header;
grant select on gv_$temp_space_header to SELECT_CATALOG_ROLE;