月归档：六月 2012

resmgr:cpu quantum等待

发表于 2012 年 6 月 14 日 由 惜分飞

昨天晚上数据库升级(使用exp/imp从9i升级到11g),开启业务,数据库出现很多resmgr:cpu quantum等待

SQL> select * from v$version;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.2.0 - 64bit Production
PL/SQL Release 11.2.0.2.0 - Production
CORE    11.2.0.2.0      Production
TNS for IBM/AIX RISC System/6000: Version 11.2.0.2.0 - Production
NLSRTL Version 11.2.0.2.0 - Production

SQL> SELECT sid,event FROM v$session WHERE wait_class#<>6;

       SID EVENT
---------- ----------------------------------------------------------------
        27 resmgr:cpu quantum
        38 resmgr:cpu quantum
        43 resmgr:cpu quantum
        46 resmgr:cpu quantum
       113 resmgr:cpu quantum
       118 resmgr:cpu quantum
       125 resmgr:cpu quantum
       140 resmgr:cpu quantum
       143 resmgr:cpu quantum
       199 resmgr:cpu quantum
       205 resmgr:cpu quantum

       SID EVENT
---------- ----------------------------------------------------------------
       212 resmgr:cpu quantum
       220 resmgr:cpu quantum
       221 resmgr:cpu quantum
       223 resmgr:cpu quantum
       238 resmgr:cpu quantum
       241 resmgr:cpu quantum
       301 resmgr:cpu quantum
       313 resmgr:cpu quantum
       314 resmgr:cpu quantum
       405 resmgr:cpu quantum
       410 resmgr:cpu quantum

       SID EVENT
---------- ----------------------------------------------------------------
       415 resmgr:cpu quantum
       435 resmgr:cpu quantum
       502 resmgr:cpu quantum
       503 resmgr:cpu quantum
       509 resmgr:cpu quantum
       510 resmgr:cpu quantum
       512 resmgr:cpu quantum
       521 resmgr:cpu quantum
       526 resmgr:cpu quantum
       528 resmgr:cpu quantum
       532 resmgr:cpu quantum

       SID EVENT
---------- ----------------------------------------------------------------
       533 enq: TX - row lock contention
       589 resmgr:cpu quantum
       596 resmgr:cpu quantum
       600 resmgr:cpu quantum
       609 resmgr:cpu quantum
       611 resmgr:cpu quantum
       625 resmgr:cpu quantum
       635 null event
       707 resmgr:cpu quantum
       727 resmgr:cpu quantum
       731 SQL*Net message to client

44 rows selected.

查询alert日志

Sat Jun 09 06:00:00 2012
Setting Resource Manager plan SCHEDULER[0x310C]:DEFAULT_MAINTENANCE_PLAN via scheduler window
Setting Resource Manager plan DEFAULT_MAINTENANCE_PLAN via parameter
Sun Jun 10 02:00:00 2012
Closing Resource Manager plan via scheduler window
Clearing Resource Manager plan via parameter

Sun Jun 10 06:00:00 2012
Setting Resource Manager plan DEFAULT_MAINTENANCE_PLAN via parameter
Mon Jun 11 02:00:00 2012
Closing Resource Manager plan via scheduler window
Clearing Resource Manager plan via parameter

Mon Jun 11 22:00:00 2012
Setting Resource Manager plan DEFAULT_MAINTENANCE_PLAN via parameter
Tue Jun 12 02:00:00 2012
Closing scheduler window
Closing Resource Manager plan via scheduler window
Clearing Resource Manager plan via parameter

Tue Jun 12 22:00:00 2012
Setting Resource Manager plan SCHEDULER[0x3108]:DEFAULT_MAINTENANCE_PLAN via scheduler window
Setting Resource Manager plan DEFAULT_MAINTENANCE_PLAN via parameter
Wed Jun 13 02:00:00 2012
Closing scheduler window
Closing Resource Manager plan via scheduler window
Clearing Resource Manager plan via parameter

从这里可以看出来，因为SCHEDULER定时启动和关闭资源管理的DEFAULT_MAINTENANCE_PLAN从而导致在晚上10点到2点Resource Manager plan处于启用状态.上线测试刚好在晚上2点之前,所有当时查询的时候发现很多resmgr:cpu quantum等待是因为Resource Manager plan启用导致(使用SCHEDULER控制其启用和关闭),很多情况下数据库跑的应用比较单一，不是十分的需要启动资源管理.
在11g中关闭方法如下

1. Set the current resource manager plan to null (or another plan that is not restrictive):

alter system set resource_manager_plan='' scope=both;

2. Change the active windows to use the null resource manager plan (or other nonrestrictive plan) using:

execute dbms_scheduler.set_attribute('WEEKNIGHT_WINDOW','RESOURCE_PLAN',''); 
execute dbms_scheduler.set_attribute('WEEKEND_WINDOW','RESOURCE_PLAN','');

3. Then, for each window_name (WINDOW_NAME from DBA_SCHEDULER_WINDOWS), run:

execute dbms_scheduler.set_attribute('<window name>','RESOURCE_PLAN','');

SQL> select WINDOW_NAME  from DBA_SCHEDULER_WINDOWS;

WINDOW_NAME
------------------------------
MONDAY_WINDOW
TUESDAY_WINDOW
WEDNESDAY_WINDOW
THURSDAY_WINDOW
FRIDAY_WINDOW
SATURDAY_WINDOW
SUNDAY_WINDOW
WEEKNIGHT_WINDOW
WEEKEND_WINDOW

9 rows selected.
发表在 Oracle性能优化 | 标签为 | 4 条评论

OPTIMIZER_INDEX_CACHING和OPTIMIZER_INDEX_COST_ADJ参数说明

发表于 2012 年 6 月 14 日 由 惜分飞

OPTIMIZER_INDEX_COST_ADJ参数说明
OPTIMIZER_INDEX_COST_ADJ lets you tune optimizer behavior for access path selection to be more or less index friendly—that is, to make the optimizer more or less prone to selecting an index access path over a full table scan.
The default for this parameter is 100 percent, at which the optimizer evaluates index access paths at the regular cost. Any other value makes the optimizer evaluate the access path at that percentage of the regular cost. For example, a setting of 50 makes the index access path look half as expensive as normal.
这个数反映执行多块IO（全表扫描）的成本与执行单个IO（索引读取）的成本。保持为100，则多块IO与单块IO成本相同。设为50优化程序认为访问单块IO的成本为多块IO的一半。

OPTIMIZER_INDEX_COST_ADJ参数试验

SQL> create table t_xifenfei
  2  as
  3   select object_id,object_name from dba_objects where rownum<101;

Table created.

SQL> create index ind_t_xifenfei on t_xifenfei(object_id);

Index created.

SQL> EXEC DBMS_STATS.gather_table_stats(user,'T_XIFENFEI',CASCADE=>TRUE);

PL/SQL procedure successfully completed.

SQL>  show parameter OPTIMIZER_INDEX;

NAME                                 TYPE                   VALUE
------------------------------------ ---------------------- ------------------------------
optimizer_index_caching              integer                0
optimizer_index_cost_adj             integer                100

SQL> set linesize 150
SQL> set autot trace ext
SQL> select object_name from t_xifenfei where object_id>100;


Execution Plan
----------------------------------------------------------
Plan hash value: 2444553208

----------------------------------------------------------------------------------------------
| Id  | Operation                   | Name           | Rows  | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |                |     1 |    11 |     2   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| T_XIFENFEI     |     1 |    11 |     2   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | IND_T_XIFENFEI |     1 |       |     1   (0)| 00:00:01 |
----------------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - access("OBJECT_ID">100)

SQL> alter session set OPTIMIZER_INDEX_COST_ADJ=1000;

Session altered.

SQL> select object_name from t_xifenfei where object_id>100;

Execution Plan
----------------------------------------------------------
Plan hash value: 548923532

--------------------------------------------------------------------------------
| Id  | Operation         | Name       | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |            |     1 |    11 |     3   (0)| 00:00:01 |
|*  1 |  TABLE ACCESS FULL| T_XIFENFEI |     1 |    11 |     3   (0)| 00:00:01 |
--------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   1 - filter("OBJECT_ID">100)

通过试验证明修改OPTIMIZER_INDEX_COST_ADJ会导致执行计划走index还是全表扫描

OPTIMIZER_INDEX_CACHING参数说明
OPTIMIZER_INDEX_CACHING lets you adjust the behavior of cost-based optimization to favor nested loops joins and IN-list iterators.The cost of executing an index using an IN-list iterator or of executing a nested loops join when an index is used to access the inner table depends on the caching of that index in the buffer cache. The amount of caching depends on factors that the optimizer cannot predict, such as the load on the system and the block access patterns of different users.You can modify the optimizer’s assumptions about index caching for nested loops joins and IN-list iterators by setting this parameter to a value between 0 and 100 to indicate the percentage of the index blocks the optimizer should assume are in the cache. Setting this parameter to a higher value makes nested loops joins and IN-list iterators look less expensive to the optimizer. As a result, it will be more likely to pick nested loops joins over hash or sort-merge joins and to pick indexes using IN-list iterators over other indexes or full table scans. The default for this parameter is 0, which results in default optimizer behavior.
这个表明的是在nested loops joins and IN-list iterators的时候,如果使用了OPTIMIZER_INDEX_CACHING参数,表明两个表关联的时候优化器考虑index cache的比例,从而选择不同的执行计划.而不是网上所说的优化器考虑所有情况下的index的cache情况(这个参数只有在nested loops joins and IN-list iterators表关联的时候的index才会被优化器考虑[index cache的比例]).进一步说明:这个参数影响两个表关联的时候是选择hash jion还是nested loops joins/sort-merge joins

总结说明
1.关于OPTIMIZER_INDEX_CACHING的参数效果我未试验出来(可能方法不正确)
2.根据网上建议在oltp系统中设置
optimizer_index_caching = 0 optimizer_index_cost_adj = 100 的默认值，一般时候数据仓库报表系统。
optimizer_index_caching = 90 optimizer_index_cost_adj = 25-50 一般时候事务处理/OLTP系统
3.设置这些参数可能存在bug

发表在 Oracle | 一条评论

因IPC导致多个监听不能正常启动

发表于 2012 年 6 月 12 日 由 惜分飞

在一台机器上装了9.2.0.4和10.1.0.4数据库,启动监听时候发现有一个启动不了
9i监听配置

LISTENER =
  (DESCRIPTION_LIST =
    (DESCRIPTION =
      (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = TCP)(HOST = xifenfei.com)(PORT = 1522))
      )
      (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC))
      )
    )
  )

SID_LIST_LISTENER =
  (SID_LIST =
    (SID_DESC =
      (GLOBAL_DBNAME = xifenfei)
      (ORACLE_HOME = /u01/oracle/9.2.0/db_1)
      (SID_NAME = xff)
    )
  )

10g监听配置

SID_LIST_LISTENER =
  (SID_LIST =
    (SID_DESC =
      (SID_NAME = PLSExtProc)
      (ORACLE_HOME = /u02/app/oracle/db10g)
      (PROGRAM = extproc)
    )
  )

LISTENER =
  (DESCRIPTION_LIST =
    (DESCRIPTION =
      (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC))
      )
      (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = TCP)(HOST = xifenfei.com)(PORT = 1521))
      )
    )
  )

分别启动监听
启动10g监听

[oraem@xifenfei ~]$ lsnrctl start

LSNRCTL for Linux: Version 10.1.0.4.0 - Production on 12-JUN-2012 15:28:51

Copyright (c) 1991, 2004, Oracle.  All rights reserved.

Starting /u02/app/oracle/db10g/bin/tnslsnr: please wait...

TNSLSNR for Linux: Version 10.1.0.4.0 - Production
System parameter file is /u02/app/oracle/db10g/network/admin/listener.ora
Log messages written to /u02/app/oracle/db10g/network/log/listener.log
Listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=ipc)(KEY=EXTPROC)))
Listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=xifenfei.com)(PORT=1521)))

Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=IPC)(KEY=EXTPROC)))
STATUS of the LISTENER
------------------------
Alias                     LISTENER
Version                   TNSLSNR for Linux: Version 10.1.0.4.0 - Production
Start Date                12-JUN-2012 15:28:51
Uptime                    0 days 0 hr. 0 min. 0 sec
Trace Level               off
Security                  ON: Local OS Authentication
SNMP                      OFF
Listener Parameter File   /u02/app/oracle/db10g/network/admin/listener.ora
Listener Log File         /u02/app/oracle/db10g/network/log/listener.log
Listening Endpoints Summary...
  (DESCRIPTION=(ADDRESS=(PROTOCOL=ipc)(KEY=EXTPROC)))
  (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=xifenfei.com)(PORT=1521)))
Services Summary...
Service "PLSExtProc" has 1 instance(s).
  Instance "PLSExtProc", status UNKNOWN, has 1 handler(s) for this service...
The command completed successfully

启动9i监听

[oracle@xifenfei ~]$ lsnrctl start

LSNRCTL for Linux: Version 9.2.0.4.0 - Production on 12-JUN-2012 15:01:44

Copyright (c) 1991, 2002, Oracle Corporation.  All rights reserved.

Starting /u01/oracle/9.2.0/db_1/bin/tnslsnr: please wait...

TNSLSNR for Linux: Version 9.2.0.4.0 - Production
System parameter file is /u01/oracle/9.2.0/db_1/network/admin/listener.ora
Log messages written to /u01/oracle/9.2.0/db_1/network/log/listener.log
Listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=xifenfei.com)(PORT=1522)))
Error listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=IPC)(KEY=EXTPROC)))
TNS-12542: TNS:address already in use
 TNS-12560: TNS:protocol adapter error
  TNS-00512: Address already in use
   Linux Error: 98: Address already in use

Listener failed to start. See the error message(s) above...

这里很奇怪两个监听使用不同的端口,为什么不能提示Address被占用呢,难道9i的已经启动了

查看9i的监听状态

[oracle@xifenfei ~]$ lsnrctl status

LSNRCTL for Linux: Version 9.2.0.4.0 - Production on 12-JUN-2012 15:02:23

Copyright (c) 1991, 2002, Oracle Corporation.  All rights reserved.

Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=xifenfei)(PORT=1522)))
TNS-12541: TNS:no listener
 TNS-12560: TNS:protocol adapter error
  TNS-00511: No listener
   Linux Error: 111: Connection refused
Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=IPC)(KEY=EXTPROC)))
TNS-12618: TNS:versions are incompatible

9i监听不能启动原因
这里有重大发现TNS:versions are incompatible.我这台服务器只有一个9i和一个10g的数据库出现版本不兼容,那就是说这个命令可能访问了10g的库中的每个监听信息.继续往上看,发现(DESCRIPTION=(ADDRESS=(PROTOCOL=IPC)(KEY=EXTPROC))).
When a process is on the same machine as the server, use the IPC protocol for connectivity instead of TCP. Inner Process Communication on the same machine does not have the overhead of packet building and deciphering that TCP has.
通过这段话可以看出IPC(Inner Process Communication)是使用于本机的内部通讯,不用包的封装,可以很大程度上提高程序执行效率.看到上面的IPC提示,我们可以确定该错误是因为10g中已经启用了IPC,然后9i的监听也要来启动这个,从而导致该错误.

解决办法
因为在同一个机器上,只能其中的一个数据库启用IPC.修改9i的监听配置

LISTENER =
  (DESCRIPTION_LIST =
    (DESCRIPTION =
      (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = TCP)(HOST = localhost)(PORT = 1522))
      )
    #  (ADDRESS_LIST =
    #    (ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC))
    #  )
    )
  )

SID_LIST_LISTENER =
  (SID_LIST =
    (SID_DESC =
      (SID_NAME = PLSExtProc)
      (ORACLE_HOME = /u01/oracle/9.2.0/db_1)
      (PROGRAM = extproc)
    )
    (SID_DESC =
      (GLOBAL_DBNAME = xifenfei)
      (ORACLE_HOME = /u01/oracle/9.2.0/db_1)
      (SID_NAME = xff)
    )
  )

启动9i监听

[oracle@xifenfei ~]$ lsnrctl

LSNRCTL for Linux: Version 9.2.0.4.0 - Production on 12-JUN-2012 15:12:55

Copyright (c) 1991, 2002, Oracle Corporation.  All rights reserved.

Welcome to LSNRCTL, type "help" for information.

LSNRCTL> start
Starting /u01/oracle/9.2.0/db_1/bin/tnslsnr: please wait...

TNSLSNR for Linux: Version 9.2.0.4.0 - Production
System parameter file is /u01/oracle/9.2.0/db_1/network/admin/listener.ora
Log messages written to /u01/oracle/9.2.0/db_1/network/log/listener.log
Listening on: (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=xifenfei.com)(PORT=1522)))

Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=localhost)(PORT=1522)))
STATUS of the LISTENER
------------------------
Alias                     LISTENER
Version                   TNSLSNR for Linux: Version 9.2.0.4.0 - Production
Start Date                12-JUN-2012 15:12:57
Uptime                    0 days 0 hr. 0 min. 0 sec
Trace Level               off
Security                  OFF
SNMP                      OFF
Listener Parameter File   /u01/oracle/9.2.0/db_1/network/admin/listener.ora
Listener Log File         /u01/oracle/9.2.0/db_1/network/log/listener.log
Listening Endpoints Summary...
  (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=xifenfei.com)(PORT=1522)))
Services Summary...
Service "PLSExtProc" has 1 instance(s).
  Instance "PLSExtProc", status UNKNOWN, has 1 handler(s) for this service...
Service "xifenfei" has 1 instance(s).
  Instance "xff", status UNKNOWN, has 1 handler(s) for this service...
The command completed successfully

查看两个监听进程

[oracle@xifenfei ~]$ ps -ef|grep tns|grep -v grep
oraem    18099     1  0 13:27 ?        00:00:00 /u02/app/oracle/db10g/bin/tnslsnr LISTENER -inherit
oracle   24312     1  0 15:12 pts/0    00:00:00 /u01/oracle/9.2.0/db_1/bin/tnslsnr LISTENER -inherit
发表在 Oracle 监听 | 一条评论