01 性能优化基础怀实践 之 ASH分析

1、模拟一个会话阻塞的场景。

   通过update 同一行数据达到模拟阻塞的效果 ：

SQL> create table t1 (id number ,name varchar2(20)) ; 
SQL> insert into t1 values(1,'huyi') ;
已创建 1 行。
SQL> insert into t1 values(2,'zhanges');
已创建 1 行。


SQL> select * from scott.t1 ;

        ID NAME
---------- --------------------
         1 huyi
         2 zhanges
SQL>  update  scott.t1 set name='huyi2'  where id=1; 

已更新 1 行。

注意：未提交         
         
新开一个窗口：
ora11g-> sqlplus "/ as sysdba"

SQL*Plus: Release 11.2.0.4.0 Production on 星期五 10月 19 08:16:26 2018

Copyright (c) 1982, 2013, Oracle.  All rights reserved.

连接到: 
Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options

SQL>  update  scott.t1 set name='huyi3'  where id=1;   产生等待

33

 

1

SQL> create table t1 (id number ,name varchar2(20)) ; 

2

SQL> insert into t1 values(1,'huyi') ;

3

已创建 1 行。

4

SQL> insert into t1 values(2,'zhanges');

5

已创建 1 行。

6

7

8

SQL> select * from scott.t1 ;

9

10

        ID NAME

11

---------- --------------------

12

         1 huyi

13

         2 zhanges

14

SQL>  update  scott.t1 set name='huyi2'  where id=1; 

15

16

已更新 1 行。

17

18

注意：未提交         

19

20

新开一个窗口：

21

ora11g-> sqlplus "/ as sysdba"

22

23

SQL*Plus: Release 11.2.0.4.0 Production on 星期五 10月 19 08:16:26 2018

24

25

Copyright (c) 1982, 2013, Oracle.  All rights reserved.

26

27

连接到: 

28

Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production

29

With the Partitioning, OLAP, Data Mining and Real Application Testing options

30

31

SQL>  update  scott.t1 set name='huyi3'  where id=1;   产生等待

32

33

2、利用ASH视图分析，定位到阻塞源会话，并解除阻塞。

      分思路：1.找到某个大至时间范围 即：什么时间

                      2.找到这个时间段的等待事件：产生原因 

                      3.找到是谁导致了阻塞 ： 阻塞源

                      4.源会话分析 

                      5.阻塞关系分析

SQL> alter session set NLS_DATE_FORMAT='YYYYMMDD HH24:MI:SS'  ;
会话已更改。

1.找到某个大至时间范围 即：什么时间
SQL> select trunc(sample_time,'mi'),count(1) from gv$active_session_history group by trunc(sample_time,'mi') order by 1 ;
TRUNC(SAMPLE_TIME   COUNT(1)
----------------- ----------

20181019 07:56:00 1
20181019 07:57:00 3
20181019 07:58:00 1
20181019 08:00:00 3
20181019 08:01:00 1
20181019 08:03:00 1
20181019 08:06:00 2
20181019 08:09:00 3
20181019 08:11:00 2
20181019 08:13:00 1
20181019 08:15:00 2
20181019 08:17:00 59  #这里等待时事明显突然增多   
20181019 08:18:00 64
20181019 08:19:00 63
20181019 08:20:00 60
20181019 08:21:00 62
20181019 08:22:00 60
20181019 08:23:00 60
20181019 08:24:00 61
20181019 08:25:00 60
20181019 08:26:00 61
20181019 08:27:00 62

2. 找出某个等待事件 ：

SQL> select trunc(sample_time,'mi'),event,count(1) from  gv$active_session_history 
  2  where 
  3   sample_time >=to_date('20181019 08:14:00','yyyymmdd hh24:mi:ss')
  4   and  sample_time <=to_date('20181019 08:18:00','yyyymmdd hh24:mi:ss')
  5   group by  trunc(sample_time,'mi') ,event order by 1 ;  

TRUNC(SAMPLE_TIME EVENT                                 COUNT(1)
----------------- ----------------------------------- ----------
20181019 08:15:00                                              1
20181019 08:15:00 log file parallel write                      1
20181019 08:17:00 enq: TX - row lock contention               59        #根据时间范围找出等待事件

3. 找出会话源头：

col sample_time for a20
col final_block for 99999999
col sid_chain for a15
col event_chain for a35


with ash as (select instance_number,session_id,event,blocking_session,program,to_char(sample_time,'YYYYMMDD HH24MISS') SAMPLE_TIME,sample_id,blocking_inst_id 
from dba_hist_active_sess_history
where sample_time >to_date('20181019 08:14:00','yyyymmdd hh24:mi:ss')
 and  sample_time <to_date('20181019 08:18:00','yyyymmdd hh24:mi:ss')
 )
 select * from (
 select sample_time,blocking_session final_block,sys_connect_by_path(session_id,',') sid_chain,
 sys_connect_by_path(event,',') event_chain
 from ash start with session_id is not null
 connect by prior blocking_session=session_id and 
 prior instance_number=blocking_inst_id and sample_id = prior sample_id) a
 where instr(sid_chain,final_block)=0 and not exists
 (select 1 from ash  b where a.final_block=b.session_id and b.blocking_session is not null)
 order by sample_time ;

SAMPLE_TIME          FINAL_BLOCK SID_CHAIN       EVENT_CHAIN
-------------------- ----------- --------------- -----------------------------------
20181019 081703              129 ,130            ,enq: TX - row lock contention       #从这里可以看出是sid 129 阻赛了130
20181019 081713              129 ,130            ,enq: TX - row lock contention
20181019 081723              129 ,130            ,enq: TX - row lock contention
20181019 081733              129 ,130            ,enq: TX - row lock contention
20181019 081743              129 ,130            ,enq: TX - row lock contention
20181019 081753              129 ,130            ,enq: TX - row lock contention

已选择6行。

4.源会话分析  
找出某个SQL引起的等待事件
select event,sql_id,count(1) from  gv$active_session_history  a where  a.sample_time > sysdate -15/144 group by event,sql_id order by count(1) desc ; 

EVENT                               SQL_ID            COUNT(1)
----------------------------------- --------------- ----------
enq: TX - row lock contention       gw47xzmm9h6qq         5921
control file parallel write                                 34
log file parallel write                                      5
os thread startup                                            2
db file sequential read             fsk02m4793dqq            1
null event                          3ktacv9r56b51            1
                                    dayq182sk41ks            1
                                    f318xdxdn0pdc            1
null event                                                   1
                                    5d2u345mmudn9            1
                                    1np800rjbv1kn            1


5.查询会话状态　

set linesize 2000
set pagesize 2000
col ins_sid for a15
col username for a10
col machine for a10
col program for a26
col module for a26
col action for a10
col sql_id for a15
col event for a25
col blocking_session for 999999999
col logon_time for a26
col prev_exec_start for a10
col client_info for a10

select inst_id||','||sid ins_sid,username,machine,program,module,action,sql_id,event,blocking_session,logon_time,prev_exec_start,client_info from gv$session where status='ACTIVE' and type<>'BACKGROUND' order by inst_id,sid;

INS_SID   USERNAME   MACHINE    PROGRAM                    MODULE                 ACTION     SQL_ID          EVENT                 BLOCKING_SESSION LOGON_TIME           PREV_EXEC_ CLIENT_INF
----------- ------ ---------- -------------------------- -------------------------- ------ --------------- ------------------------- ------------ -------------------- ---------- ----------
1,8         SYS        ora11g     sqlplus@ora11g (TNS V1-V3) sqlplus@ora11g (TNS V1-V3)        c3wctdc71swjh   SQL*Net message to client              20181019 08:24:18      20181019 0
                                                                                                                                                                             9:49:55

1,70                   ora11g     oracle@ora11g (J000)                                                         jobq slave wait                        20181019 09:49:46
1,130       SYS        ora11g     sqlplus@ora11g (TNS V1-V3) sqlplus@ora11g (TNS V1-V3)        gw47xzmm9h6qq   enq: TX - row lock conten          129 20181019 08:16:26      20181019 0
                                                                                                               tion                                                          8:16:26

1,131                  ora11g     oracle@ora11g (J001)                                                             jobq slave wait                            20181019 09:49:46


查询被阻塞的SQL语句：
select SQL_ID,sql_text from v$sqltext where SQL_ID='&sql_id' order by piece asc;   
SQL_ID          SQL_TEXT
--------------- ----------------------------------------------------------------------------------------------------
gw47xzmm9h6qq    update  scott.t1 set name='huyi3'  where id=1;

查询阻塞的SQL语句：
set linesize 1000 
select /*+ ORDERED USE_NL(st) */ INST_ID,SQL_ID,sql_text from gv$sqltext a 
where (a.hash_value, a.address) IN (
            SELECT DECODE (sql_hash_value,0, prev_hash_value,sql_hash_value),
                   DECODE (sql_hash_value, 0,prev_sql_addr, sql_address)  
            from gv$session where sid=&sid and INST_ID=INST_ID ) order by piece asc;
输入 sid 的值:  129
原值    5:             from gv$session where sid=&sid and INST_ID=INST_ID ) order by piece asc
新值    5:             from gv$session where sid=129 and INST_ID=INST_ID ) order by piece asc

   INST_ID SQL_ID          SQL_TEXT
---------- --------------- ----------------------------------------------------------------------------------------------------
         1 4f0caa4cy9x8v    update  scott.t1 set name='huyi2'  where id=1

#可以看出：是由于会话129 更新了一条数据，未提交，导到了另一个会话更新相同数据时被阻塞。  
#    总结:1.出现类似问题思路清晰，找到时间段，是什么原因，出现什么问题　，理清问题关系
#　　　　 2.要深度了解gv$active_session_history　的数据源，以及各字段的含义，以方便我们能没在有工具的情况下，编写出我们需要的脚本。


历史等待事件状态：
set linesize 2000
set pagesize 2000
col event for a35

select event,count(1) from gv$active_session_history a where a.sample_time > sysdate -15/144 group by event order by count(1) desc ;
EVENT                                 COUNT(1)
----------------------------------- ----------
enq: TX - row lock contention             5871
                                            71
control file parallel write                 34
log file parallel write                      5
null event                                   2
os thread startup                            2
db file sequential read                      1

已选择7行。

173

 

1

SQL> alter session set NLS_DATE_FORMAT='YYYYMMDD HH24:MI:SS'  ;

2

会话已更改。

3

4

1.找到某个大至时间范围 即：什么时间

5

SQL> select trunc(sample_time,'mi'),count(1) from gv$active_session_history group by trunc(sample_time,'mi') order by 1 ;

6

TRUNC(SAMPLE_TIME   COUNT(1)

7

----------------- ----------

8

9

20181019 07:56:00 1

10

20181019 07:57:00 3

11

20181019 07:58:00 1

12

20181019 08:00:00 3

13

20181019 08:01:00 1

14

20181019 08:03:00 1

15

20181019 08:06:00 2

16

20181019 08:09:00 3

17

20181019 08:11:00 2

18

20181019 08:13:00 1

19

20181019 08:15:00 2

20

20181019 08:17:00 59  #这里等待时事明显突然增多   

21

20181019 08:18:00 64

22

20181019 08:19:00 63

23

20181019 08:20:00 60

24

20181019 08:21:00 62

25

20181019 08:22:00 60

26

20181019 08:23:00 60

27

20181019 08:24:00 61

28

20181019 08:25:00 60

29

20181019 08:26:00 61

30

20181019 08:27:00 62

31

32

2. 找出某个等待事件 ：

33

34

SQL> select trunc(sample_time,'mi'),event,count(1) from  gv$active_session_history 

35

  2  where 

36

  3   sample_time >=to_date('20181019 08:14:00','yyyymmdd hh24:mi:ss')

37

  4   and  sample_time <=to_date('20181019 08:18:00','yyyymmdd hh24:mi:ss')

38

  5   group by  trunc(sample_time,'mi') ,event order by 1 ;  

39

40

TRUNC(SAMPLE_TIME EVENT                                 COUNT(1)

41

----------------- ----------------------------------- ----------

42

20181019 08:15:00                                              1

43

20181019 08:15:00 log file parallel write                      1

分析思路：

来自为知笔记(Wiz)