DBA之路：OoNiceDream

What is ORACLE_HOME_NAME

OoNiceDream — Fri, 21 May 2010 03:31:41 +0000

ORACLE_HOME_NAME 其实跟ORACLE_HOME的含义是一样的，是在oracle初始版本安装时指定的。
在最近一次升级9208时发现，指定补丁安装的目录名称时，这个ORACLE_HOME_NAME一定要跟初始安装时一样，否则就会报错，例如这里我写成了OUIHome1：

这里提示的oradb就是ORACLE_HOME_NAME，太有误导性了，搞成跟数据库名一样，半天没看明白。

要确认ORACLE_HOME_NAME，还可以通过oraInventory目录下的ContentsXML/inventory.xml文件：

# cat /etc/oraInst.loc
inventory_loc=/oradata/oraInventory
inst_group=oinstall
# cd /oradata/oraInventory
# cd ContentsXML
# cat inventory.xml
< ?xml version="1.0" standalone="yes" ?>
<!-- Copyright (c) 2001 Oracle Corporation. All rights Reserved -->
<!-- Do not modify the contents of this file by hand. -->
<inventory>
<version_info>
   <saved_with>2.2.0.18.0saved_with>
   <minimum_ver>2.1.0.6.0minimum_ver>
version_info>
<home_list>
<home NAME="oradb" LOC="/oracle/product/9.2.0.4" TYPE="O" IDX="1"/>
home_list>
inventory>

这里的HOME NAME指的就是ORACLE_HOME_NAME，而oradb就是ORACLE_HOME_NAME的value了。

— The End —

Linux下OUI乱码

OoNiceDream — Wed, 19 May 2010 01:48:20 +0000

在Linux下遇到OUI乱码，第一次遇到这样的情况，首先想到环境变量设置不对。
设置以下环境变量后，未再显示乱码：

export LC_ALL=en_US

后来同事也提供了一种方法：

unset LANG

查找了一些资料：点击查看相关文章

LC_ALL与LANG都是与语言相关的环境变量的统称，其它还有很多，比如：

LC _CTYPE, LC_TIME, LC_NUMERIC, LC_COLLATE

其中LC_ALL优先级最高，而LANG优先级最低。

— The End —

RAC环境下library cache pin/lock的处理

OoNiceDream — Mon, 17 May 2010 02:50:52 +0000

这个案例放着很久了，最近又有遇到，小结了一下。
RAC环境下，主要是不同实例上handle地址（kglpnhdl or kgllkhdl）不同，所以如果pin/lock被不同的节点持有时，处理起来就略有不同，看下面这个例子：

实例1上进行授权操作，发现会话挂起，等待事件为：library cache pin:

SQL> select inst_id, sid, event, p1raw
  2    from gv$session_wait
  3   where event = 'library cache pin';
 
   INST_ID        SID EVENT                P1RAW
---------- ---------- -------------------- ----------------
         1       2325 library cache pin    C00000087F12E1F8

实例1上，只有一个请求pin的会话，也就是当前在做授权的会话：

SQL> select inst_id, KGLPNUSE, KGLPNHDL, KGLPNMOD, KGLPNREQ
  2    from x$kglpn
  3   where kglpnhdl = 'C00000087F12E1F8'
  4   order by KGLPNREQ;
 
INST_ID KGLPNUSE         KGLPNHDL           KGLPNMOD   KGLPNREQ
------- ---------------- ---------------- ---------- ----------
      1 C0000007C97625C0 C00000087F12E1F8          0          3
 
SQL> select sid,serial#,sql_hash_value,username,machine from v$session
  2  where saddr='C0000007C97625C0';
 
   SID    SERIAL# SQL_HASH_VALUE USERNAME  MACHINE
------ ---------- -------------- --------- -------
  2325       5103     3038871805 SYS       host1
 
SQL> select kglnaown, kglnaobj
  2    from X$KGLOB
  3   where KGLHDADR = 'C00000087F12E1F8';
 
KGLNAOWN   KGLNAOBJ
---------- ---------------
U_NAME     OBJ2CHANNEL

那么这个对象很有可能在实例2上被pin住了。
每个实例都有自己的shared pool，handle地址不同也比较好理解，所以不能用handle地址去查找了。
但有一点是不会变的，那就是x$kglob中显示的被pin住的对象，因此，这个handle地址就可以从x$kglob中获取了。
几个视图关联下，就可以通过以下这个语句在另一个节点上来查找持有pin的会话：

SQL> select p.inst_id,
  2         s.sid,
  3         s.sql_hash_value,
  4         s.username,
  5         s.osuser,
  6         s.machine,
  7         p.KGLPNUSE,
  8         p.KGLPNHDL,
  9         p.KGLPNMOD,
 10         p.KGLPNREQ,
 11         o.kglnaobj
 12    from v$session s, x$kglpn p, x$kglob o
 13   where o.kglnaobj = 'OBJ2CHANNEL'
 14     and p.kglpnhdl = o.kglhdadr
 15     and s.SADDR = p.kglpnuse
 16     and p.KGLPNMOD <> 0;
 
INST_ID SID SQL_HASH_VALUE USERNAME OSUSER MACHINE KGLPNUSE         KGLPNHDL         KGLPNMOD KGLPNREQ KGLNAOBJ
------- --- -------------- -------- ------ ------- ---------------- ---------------- -------- -------- ---------
      2 197     4062951240 U_NAME   u_name host2   C00000081E3E5C60 C000000888817D68        2        0 OBJ2CHANN
      2 197     4062951240 U_NAME   u_name host2   C00000081E3E5C60 C00000088ED88980        2        0 OBJ2CHANN

library cache lock处理类同，只是视图换一下（x$kgllk）。
— The End —

HA与RAC组件

OoNiceDream — Sat, 15 May 2010 14:47:47 +0000

最近在AIX下重新安装数据库软件时，发现在已安装HA的情况下，默认总是会安装RAC组件，执行rootpre.sh时会有如下提示：

# ./rootpre.sh
 
Configuring Asynchronous I/O...
Asynchronous I/O is already defined
/bin/chgrp: unknown group: hagsuser
Unable to change the group ownership of /var/ha/soc/grpsvcsdsocket.SXCZ_cluster.
Aborting pre-installation procedure. Installations of Oracle may fail.

在rootpre.sh中，找到判断HA的这行：

isHACMP=`/bin/odmget HACMPnode 2>/dev/null | wc -l`

改了一下，未再要求添加hagsuser组，但安装时，仍会安装RAC组件。

根据以前的一些笔记，在这样的情况下，创建数据选择非cluster database时，会报ORA-32700错误：

$ oerr ora 32700 
32700, 00000, "error occurred in DIAG Group Service"
// *Cause: An unexpected error occurred while performing a DIAG Group Service
//         operation.
// *Action: Verify that the DIAG process is still active. Also, check the
//          Oracle DIAG trace files for errors.

没找到取消RAC组件的安装办法，最终还是在软件安装完成后，再卸载掉RAC组件：

make -f ins_rdbms.mk rac_off
make -f ins_rdbms.mk ioracle

— The End —

西塘印象

OoNiceDream — Tue, 30 Mar 2010 14:16:07 +0000

上周六去西塘逛了一下，其实我想说的是，西塘这地方，不咋好玩。

总体印象：人多、水脏，街道窄。

估计坐下来喝杯茶，比逛来逛去，感觉要好。

晚上夜景应该会好一点，不过我们没呆太久。

— The End —

AIX 6.1下RAC安装问题小结

OoNiceDream — Tue, 30 Mar 2010 06:00:00 +0000

最近在AIX 6.1上安装了一套RAC，实施得不多，遇到不少问题，记录一下：

OS版本：

HOST_NAM_1:/#oslevel -s
6100-04-02-1007

HA版本：

HOST_NAM_1:/#lslpp -l cluster.*
  Fileset                      Level  State      Description         
  ----------------------------------------------------------------------------
Path: /usr/lib/objrepos
  cluster.adt.es.client.include
                             5.5.0.0  COMMITTED  ES Client Include Files
  cluster.adt.es.client.samples.clinfo
                             5.5.0.0  COMMITTED  ES Client CLINFO Samples
  cluster.adt.es.client.samples.clstat
                             5.5.0.1  COMMITTED  ES Client Clstat Samples

### RSH 报错

#rsh HOST_NAM_2 date
rshd: 0826-813 Permission is denied.

相关文件配置：

#cat .rhosts
HOST_NAM_1 root
HOST_NAM_2 root
HOST_NAM_1 oracle
HOST_NAM_2 oracle
 
#cat /etc/hosts.equiv
HOST_NAM_1 root
HOST_NAM_2 root
HOST_NAM_1 oracle
HOST_NAM_2 oracle

其中HOST_NAM_1、HOST_NAM_2是HOSTNAME。

这里主要是/etc/hosts文件中，HOSTNAME不能当作别名，或者，”.rhosts”、”hosts.equiv”里不要配别名，应该是跟解析有关。

原HOSTS配置：

175.16.1.11    HOST_NAM_1_boot1  HOST_NAM_1
175.16.1.12    HOST_NAM_2_boot1  HOST_NAM_2
 
192.168.10.17   HOST_NAM_1_boot2
192.168.10.18   HOST_NAM_2_boot2
 
192.168.10.16   HOST_NAM_2_vip
192.168.10.15   HOST_NAM_1_vip

改为：

175.16.1.11    HOST_NAM_1_boot1  
175.16.1.12    HOST_NAM_2_boot1      
 
192.168.10.17   HOST_NAM_1  HOST_NAM_1_boot2
192.168.10.18   HOST_NAM_2  HOST_NAM_2_boot2
 
192.168.10.16   HOST_NAM_2_vip
192.168.10.15   HOST_NAM_1_vip

### rootpre.sh报错

这个安装前在文档中有到看到，作为注意事项记录一下：
The Oracle 10gR2 OUI and configuration assistant programs do not recognize AIX 6 V6.1 as a supported release.
执行rootpre.sh时会报：

Configuring Asynchronous I/O....
 
Asynchronous I/O is not installed on this system.
 
You will need to install it, and either configure it yourself using
 
'smit aio' or rerun the Oracle root installation procedure.
 
 
 
Configuring POSIX Asynchronous I/O....
 
Posix Asynchronous I/O is not installed on this system.
 
You will need to install it, and either configure it yourself using
 
'smit aio' or rerun the Oracle root installation procedure.

解决方法：下载6718715补丁，执行里面的rootpre.sh

参考文档：282036.1

### VIPCA 报错

VIPCA时，VIP起不来，日志报错信息：

Interface en4 checked failed (host=HOST_NAM_1)
Invalid parameters, or failed to bring up VIP (host=HOST_NAM_1)

原因：VIP绑定的是小机集成的网卡　Logical Host Ethernet Port (lp-hea)
The entstat output for LHEA is different from a regular adapter

解决方法：
修改racgvip脚本，找到
$ENTSTAT -d $_IF　这行，修改为：

$ENTSTAT -d $_IF | $GREP -iEq '.*lan.*state.*:.*operational.*|.*link.*status.*:.*up.*|.*port.*operational.*state.*:.*up.*|.*driver.*flags.*:.*up.*'

参考文档：959746.1

### ONS 起不来

日志报错信息：

Failed to get IP for localhost (0)
Failed to get IP for localhost (0)
Failed to get IP for localhost (0)
onsctl: ons failed to start

解决方法：
原hosts文件中找不到localhost：

127.0.0.1    loopback

改为：

127.0.0.1    loopback localhost

### CRS 升级10.2.0.4报错

升级完成后执行root102.sh，报：

# ./root102.sh
Error : Please change the CRS_ORACLE_USER id oracle 
         to have the following OS capabilities : 
  < CAP_PROPAGATE CAP_BYPASS_RAC_VMM CAP_NUMA_ATTACH >

解决方法：

#chuser capabilities=CAP_BYPASS_RAC_VMM,CAP_PROPAGATE,CAP_NUMA_ATTACH oracle
#lsuser -f oracle | grep capabilities
        capabilities=CAP_BYPASS_RAC_VMM,CAP_PROPAGATE,CAP_NUMA_ATTACH

这个报错之前遇到过，升级的文档中，也有提到。

### CRS 升级10.2.0.4后，VIP起不来

这次的有点难搞，日志中没有太多的信息，只有一行：

Invalid parameters, or failed to bring up VIP (host=HOST_NAM_1)

后来使用crsctl对VIP进行debug，收集更多的信息：

#crsctl debug log res "ora.host_nam_2.vip:5" 
Set Resource Debug Module: ora.host_nam_2.vip  Level: 5
#srvctl start nodeapps -n host_nam_2
CRS-0233: Resource or relatives are currently involved with another operation.
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:25 GMT+08:00 2010 [ 360824 ] Checking interface existance
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:25 GMT+08:00 2010 [ 360824 ] Calling getifbyip
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:25 GMT+08:00 2010 [ 360824 ] getifbyip:  started for 192.168.10.16
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:26 GMT+08:00 2010 [ 360824 ] getifbyip: checking if failover is happening ()
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:26 GMT+08:00 2010 [ 360824 ] getifbyip: failover is not happening ()
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:26 GMT+08:00 2010 [ 360824 ] Completed getifbyip
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:26 GMT+08:00 2010 [ 360824 ] ping_vip 192.168.10.16 started
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:26 GMT+08:00 2010 [ 360824 ] About to execute : /usr/sbin/ping  -c 1 -w 1 192.168.10.16
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] ping_vip: 192.168.10.16 is not pingable, _count = 1
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] Completed with initial interface test
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] Broadcast = 192.168.10.255
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] Interface tests
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] checkIf: start for if=en4
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] IsIfAlive: start for if=en4
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] defaultgw:  started
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] defaultgw:  completed with 192.168.10.254
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:27 GMT+08:00 2010 [ 360824 ] About to execute command: /usr/sbin/ping -S 192.168.10.18  -c 1 -w 1 192.168.10.254
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:28 GMT+08:00 2010 [ 360824 ] About to execute command: /usr/sbin/ping -S 192.168.10.18  -c 1 -w 1 192.168.10.254
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:29 GMT+08:00 2010 [ 360824 ] IsIfAlive: RX packets checked if=en4 failed
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:29 GMT+08:00 2010 [ 360824 ] Interface en4 checked failed (host=HOST_NAM_2)
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:29 GMT+08:00 2010 [ 360824 ] IsIfAlive: end for if=en4
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:29 GMT+08:00 2010 [ 360824 ] checkIf: end for if=en4
host_nam_2:ora.host_nam_2.vip:Invalid parameters, or failed to bring up VIP (host=HOST_NAM_2)
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] Checking interface existance
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] Calling getifbyip
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] getifbyip:  started for 192.168.10.16
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] getifbyip: checking if failover is happening ()
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] getifbyip: failover is not happening ()
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] Completed getifbyip
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] ping_vip 192.168.10.16 started
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:30 GMT+08:00 2010 [ 307376 ] About to execute : /usr/sbin/ping  -c 1 -w 1 192.168.10.16
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:31 GMT+08:00 2010 [ 307376 ] ping_vip: 192.168.10.16 is not pingable, _count = 1
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:31 GMT+08:00 2010 [ 307376 ] Completed with initial interface test
host_nam_2:ora.host_nam_2.vip:Thu Mar 25 13:57:31 GMT+08:00 2010 [ 307376 ] Broadcast = 192.168.10.255
CRS-1006: No more members to consider
CRS-0215: Could not start resource 'ora.host_nam_2.vip'.
CRS-0210: Could not find resource ora.host_nam_2.LISTENER_HOST_NAM_2.lsnr.

这才搜索到了相关信息：

Bug 8413088: VIP CANNOT START ON AIX 6.1 BECAUSE NETSTAT HAS A NEW COLUMN.
Bug 9157855: DURING RESTART OR WHEN ONE OF THE TWO NODE CLUSTER IS DOWN, VIP RESOURCE FAILS

这个问题，在打完CRS PSU后，也同样有可能存在，可以通过修改racvip脚本来解决：

10.2.0.4：

_O1=`$NETSTAT -n -I $_IF | $AWK "{ if (/^$_IF/) {print \\$5; exit}}"`
_O2=`$NETSTAT -n -I $_IF | $AWK "{ if (/^$_IF/) {print \\$5; exit}}"`

打完PSU后：

_O1=`$NETSTAT -n -I $_IF -p tcp | $GREP -iE ".*packets received$" | $AWK "{print \\$1; exit}"`
_O2=`$NETSTAT -n -I $_IF -p tcp | $GREP -iE ".*packets received$" | $AWK "{print \\$1; exit}"`

最终改为：

_O1=`$NETSTAT -n -I $_IF -p ip | $GREP -iE ".*packets received$" | $AWK "{print \\$1; exit}"`
_O2=`$NETSTAT -n -I $_IF -p ip | $GREP -iE ".*packets received$" | $AWK "{print \\$1; exit}"`

这个问题很杯具，花了很多时间，安装前，阅读相关文档时，我就注意到了这个BUG，两次VIP起不来，我都拿去对比，看看是不是这个BUG。结果还是没发现，一直到debug出来。

### RDBMS升10204时，报java进程没停

Oracle Universal Installer has detected that
there are processes running in the
currently selected Oracle Home. The
following processes need to be shutdown
before continuing:
java

刚开始时，还有CRS的进程，停掉CRS后，还有一个java始终过不去，用fuser查使用$ORACLE_HOME目录进程，全kill，把ps -ef | grep java 出来的进程，除了安装的进程外都杀了，还是不行。

最后在Metalink上找到了解决方法，升级前：

cd /usr/sbin/
mv fuser fuser.orig
touch /usr/sbin/fuser
chmod +x /usr/sbin/fuser

升级完成后，再改回来：

cd /usr/sbin/
cp fuser.orig fuser

这招很阴啊。。

参考文档：975597.1

### 数据库打完PSU补丁后，启CRS报错

在CRS、database都升级、打补丁完成后，启CRS、VIP等资源时，报错：

HOST_NAM_1:/#crsctl start crs
exec(): 0509-036 Cannot load program /app/oracle/product/10204/db_1/bin/crsctl.bin because of the following errors:
        0509-150   Dependent module libhasgen10.a(shr_hasgen10.o) could not be loaded.
        0509-022 Cannot load module libhasgen10.a(shr_hasgen10.o).
        0509-026 System error: A file or directory in the path name does not exist.
 
HOST_NAM_1:/app/oracle/product/10204/db_1/bin#./srvctl stop nodeapps -n host_nam_1
./srvctl[187]: %s_jreLocation%/bin/java:  not found.
 
HOST_NAM_1:/app/oracle/product/10204/crs_1/lib#srvctl stop nodeapps -n host_nam_1
/app/oracle/product/10204/db_1/bin/srvctl[187]: %s_jreLocation%/bin/java:  not found.

解决方法：改变环境变量，使从crs_/bin/目录下运行这些命令。

### DBCA时报错

DBCA时，在创建实例这步时，报错：

ORA-27504: IPC error creating OSD context
ORA-27300: OS system dependent operation:sendmsg failed with status: 59
ORA-27301: OS failure message: Message too long
ORA-27302: failure occurred  at: sskgxpsnd1
ORA-27303: additional information: MTU verification failed to send msg

原因：The problem was caused by incorrect UDP and TCP packet settings.

解决方法：修改以下参数：

no -o tcp_sendspace=262144
no -o tcp_recvspace=262144
no -o udp_sendspace=65536
no -o udp_recvspace=262144
no -o rfc1323=1

之前的参数都偏小：

tcp_sendspace 131072 
tcp_recvspace 131072

使用no -a查看参数设置

参考文档：300956.1

— The End —

SQL Profiles与语句优化

OoNiceDream — Tue, 23 Mar 2010 04:50:30 +0000

最近对10G的SQL Profiles做了一点学习。个人理解SQL Profiles可以看作是SQL语句的统计信息。只是这个统计信息对特定SQL语句才能起作用，不对会语句的对象、其它语句产生影响。

使用SQL Profiles前要用SQL Tuning Advisor收集对语句的优化建议，再根据优化建议创建SQL Profiles。

SQL Profiles 使用也比较灵活，可以在会话级、系统级应用。

语句绑定SQL Profile后，测试了下SQL Profile与Bind Peeking的关系。测试发现，Bind Peeking的特性还是会起作用。这从另一方面说明SQL Profile与OUTLINE的不同：绑定OUTLINE后，执行计划是被固化的；绑定SQL Profile后，执行计划不是不变，而是优化器在执行该语句时，会参考SQL Profile中的信息。

SQL Profiles会收集哪些信息？这点比较悬，没找到特别好的解释，不过从DBA_SQLTUNE_STATISTICS视图中也可以看到蛛丝马迹：

CREATE OR REPLACE VIEW DBA_SQLTUNE_STATISTICS AS
SELECT TASK_ID, OBJECT_ID, PARSING_SCHEMA_ID, MODULE, ACTION, ELAPSED_TIME,
         CPU_TIME, BUFFER_GETS, DISK_READS, DIRECT_WRITES,
         ROWS_PROCESSED, FETCHES, EXECUTIONS,
         END_OF_FETCH_COUNT, OPTIMIZER_COST, OPTIMIZER_ENV, COMMAND_TYPE
  FROM   wri$_adv_sqlt_statistics;

以下是SQL Profiles的使用步骤，记录一下，当做速查手册：

1、创建优化任务：

–使用SQL TEXT

DECLARE
 my_task_name VARCHAR2(50);
 my_sqltext   CLOB;
BEGIN
 my_sqltext := 'select /*+ FULL(T) */ * from t where n=:A';
 my_task_name := DBMS_SQLTUNE.CREATE_TUNING_TASK(
         sql_text    => my_sqltext,
         user_name   => 'ROCKEY',
         scope       => 'COMPREHENSIVE',
         time_limit  => 60,
         task_name   => 'rockey_sql_tuning_task_001',
         description => 'Task to tune a query on a sspecified table');
END;
/

–使用SQL ID:

DECLARE
 my_task_name VARCHAR2(50);
 my_sql_id   VARCHAR2(64);
BEGIN
 my_sql_id := 'gh991ctttx3k7';
 my_task_name := DBMS_SQLTUNE.CREATE_TUNING_TASK(
         sql_id    => my_sql_id,
         scope       => 'COMPREHENSIVE',
         time_limit  => 60,
         task_name   => 'rockey_sql_tuning_task_001',
         description => 'Task to tune a query on a specified table');
END;
/

2、执行优化任务：

BEGIN
  DBMS_SQLTUNE.EXECUTE_TUNING_TASK( task_name => 'rockey_sql_tuning_task_001');
end;
/

3、查看优化建议：

SET WRAP ON
SET LONG 10000
SET LONGCHUNKSIZE 1000
SET LINESIZE 130
SELECT DBMS_SQLTUNE.REPORT_TUNING_TASK( 'rockey_sql_tuning_task_001') from DUAL;
或者
SELECT dbms_advisor.GET_TASK_REPORT(task_name)
  FROM dba_advisor_tasks
 where task_name = 'rockey_sql_tuning_task_001';

其中dbms_advisor.GET_TASK_REPORT可用于查看ADDM REPORT。

4、创建SQL Profile:

DECLARE
 my_sqlprofile_name VARCHAR2(30);
BEGIN
 my_sqlprofile_name := DBMS_SQLTUNE.ACCEPT_SQL_PROFILE (
    task_name => 'rockey_sql_tuning_task_001',
    name      => 'my_sql_profile_001',
    category    => 'my_category');
END;
/

5、SQL语句绑字SQL Profile：
修改sqltune_category参数，可以会话级、系统级进行更改：

alter session set sqltune_category=MY_CATEGORY;

6、查看语句是否使用了SQL Profile：

select sql_text, sql_id, sql_profile, executions, plan_hash_value
  from v$sql
 where sql_profile is not null;

7、其它常用功能：
删除优化任务：

begin
dbms_sqltune.drop_tuning_task('rockey_sql_tuning_task_001');
end;
/
 
begin
dbms_advisor.delete_task('rockey_sql_tuning_task_001');
end;
/

删除SQL Profile：

BEGIN
  DBMS_SQLTUNE.DROP_SQL_PROFILE(name => 'my_sql_profile_001');
END;
/

与优化任务相关的常用视图：

select * from dba_sql_profiles;
 
select * from DBA_ADVISOR_TASKS;

— The End —

两种锁类型：SQ、JQ

OoNiceDream — Fri, 12 Mar 2010 03:31:54 +0000

最近在一个客户的系统上有时可以观察到不少的“enq: SQ - contention”等待，以前一般都按hash_value、event来group by，再按hash_value来找到相关语句，结果相关的SQL语句中没有看到sequence的影子。查看AWR报告，也找不到sequence。最终简单地查了dba_sequences视图，把cache size为0的都调大了。

后来与老大们聊起，才发现enq: SQ - contention的P2字段就是sequence的object id。以前查enqueue光注意P1了：

SQL> select name, parameter1, parameter2, parameter3
  2    from v$event_name
  3   where name = 'enq: SQ - contention';
 
NAME                           PARAMETER1      PARAMETER2      PARAMETER3
------------------------------ --------------- --------------- ---------------
enq: SQ - contention           name|mode       object #        0

另外查v$lock，type为SQ，效果是一样的，ID1就是sequence的object id：

SQL> select * from v$lock where type='SQ';
 
ADDR             KADDR             SID TY    ID1 ID2 LMODE REQUEST CTIME BLOCK
---------------- ---------------- ---- -- ------ --- ----- ------- ----- -----
C000000A09AAD688 C000000A09AAD6A8  972 SQ 371898   0     0       6     2     0
C000000A09AAD7C8 C000000A09AAD7E8 1445 SQ 371898   0     0       6     0     0
C000000A09AAE768 C000000A09AAE788 2021 SQ 371898   0     0       6     0     0
C000000A09AADFE8 C000000A09AAE008 2236 SQ 371898   0     0       6     1     0
C000000A09AAE3A8 C000000A09AAE3C8 2734 SQ 371898   0     0       6     0     0
C000000A09AAE588 C000000A09AAE5A8 3681 SQ 371898   0     0       6     0     0
C000000A09AAEF88 C000000A09AAEFA8 3748 SQ 371898   0     0       6     0     0
C000000A09AAE6C8 C000000A09AAE6E8 3785 SQ 371898   0     0       6     0     0
C000000A09AB0248 C000000A09AB0268 3867 SQ 371898   0     6       0     2     1
 
SQL> select object_id,owner,object_name,object_type from dba_objects
  2  where object_id=371898;
 
 OBJECT_ID OWNER   OBJECT_NAME       OBJECT_TYPE
---------- ------- ----------------- ------------
    371898 U_NAME  U_NAME_SEQUENCE_1 SEQUENCE
 
1 row selected.

JQ锁，是一次在查看dba_jobs_running非常慢时，查看了下视图创建语句发现的：

create or replace view dba_jobs_running as
select v.SID, v.id2 JOB, j.FAILURES,
    LAST_DATE, substr(to_char(last_date,'HH24:MI:SS'),1,8) LAST_SEC,
    THIS_DATE, substr(to_char(this_date,'HH24:MI:SS'),1,8) THIS_SEC,
    j.field1 INSTANCE
  from sys.job$ j, v$lock v
  where v.type = 'JQ' and j.job (+)= v.id2;

这里就可以看出v$lock中，JQ锁的ID2就是JOB的number了。

— The End —

What is rolling patch

OoNiceDream — Wed, 10 Mar 2010 13:21:02 +0000

所谓的rolling patch，最大的特点在于，允许RAC环境下，不同实例运行在不同的数据库版本上，当然这个不同，指的是打上rolling patch后造成的版本不一致：

This patch could be applied to the boxes in a cascaded fashion and the
different instances they could be running with different versions of
the code (the patch causes the code diff) and this is specific for certain
patches which can be applied and would not affect the SGA structures and heaps.

比较让人容易误解的是，以为这是一种online patch，可以不用停数据库，在线打。

rolling patch 还是需要停实例打的。比如在两个节点的RAC环境中，可以按这样的步骤操作：
1、停实例1；2、打补丁；3、启实例1；
4、停实例2；5、打补丁；6、启实例2。

其中有一段时间，实例1、实例2同时启着，且数据库版本不一致，但数据库是不受影响，可以有效地减少停机时间。

如何查看patch是否为rolling patch:
在补丁目录：/etc/config下，有一个inventory文件，其中有一项：

<online_rac_installable>trueonline_rac_installable>

若为true则为rolling patch

还可以使用opatch query命令：

 Patch created on 18 Apr 2008, 02:57:01 hrs PST8PDT
 Need to shutdown Oracle instances: true
 Patch is roll-backable: true
 Patch is a "Patchset Update": false
 Patch is a rolling patch: true
 Patch has sql related actions: false
 Patch is an online patch: false
 Patch is a portal patch: false
 Patch is an "auto-enabled" patch: false

Patch is a rolling patch 一项true则为rolling patch。

昨天在测试环境上打了几个rolling patch，是加”local”选项，一个节点全打完，启动，再打另一个节点。
未加”local”选项时，则按提示起停实例即可，以下是网上搜到的一个打rolling patch日志，供大家参考（原链接已不能用，用Google快照打开的，不附出处了）：

[oracle@gr7251 9004119]$ /u01/app/oracle/product/11.2.0/dbhome_1/OPatch/opatch apply
Invoking OPatch 11.1.0.6.6
Oracle Interim Patch Installer version 11.1.0.6.6
Copyright (c) 2009, Oracle Corporation. All rights reserved.
Oracle Home : /u01/app/oracle/product/11.2.0/dbhome_1
Central Inventory : /u01/app/oraInventory
from : /etc/oraInst.loc
OPatch version : 11.1.0.6.6
OUI version : 11.2.0.1.0
OUI location : /u01/app/oracle/product/11.2.0/dbhome_1/oui
Log file location : /u01/app/oracle/product/11.2.0/dbhome_1/cfgtoollogs/opatch/opatch2010-02-04_10-22-44AM.log
Patch history file: /u01/app/oracle/product/11.2.0/dbhome_1/cfgtoollogs/opatch/opatch_history.txt
ApplySession applying interim patch '9004119' to OH '/u01/app/oracle/product/11.2.0/dbhome_1'
Running prerequisite checks...
OPatch detected the node list and the local node from the inventory. OPatch will patch the local system then propagate the patch to the remote nodes.
This node is part of an Oracle Real Application Cluster.
Remote nodes: 'gr7265' 'gr7286'
Local node: 'gr7251'
Please shutdown Oracle instances running out of this ORACLE_HOME on the local system.
(Oracle Home = '/u01/app/oracle/product/11.2.0/dbhome_1')
Is the local system ready for patching? [y|n]
y
User Responded with: Y
Backing up files and inventory (not for auto-rollback) for the Oracle Home
Backing up files affected by the patch '9004119' for restore. This might take a while...
Backing up files affected by the patch '9004119' for rollback. This might take a while...
Patching component oracle.rdbms, 11.2.0.1.0...
Copying file to "/u01/app/oracle/product/11.2.0/dbhome_1/bin/racone2rac"
Copying file to "/u01/app/oracle/product/11.2.0/dbhome_1/bin/Omotion"
Copying file to "/u01/app/oracle/product/11.2.0/dbhome_1/bin/raconefix"
Copying file to "/u01/app/oracle/product/11.2.0/dbhome_1/bin/raconeinit"
Copying file to "/u01/app/oracle/product/11.2.0/dbhome_1/bin/raconestatus"
ApplySession adding interim patch '9004119' to inventory
Verifying the update...
Inventory check OK: Patch ID 9004119 is registered in Oracle Home inventory with proper meta-data.
Files check OK: Files from Patch ID 9004119 are present in Oracle Home.
The local system has been patched. You can restart Oracle instances on it.
Patching in rolling mode.
Remaining nodes to be patched:
'gr7265' 'gr7286'
What is the next node to be patched?
gr7265
You have selected 'gr7265' from 'gr7265' 'gr7286'
The node 'gr7265' will be patched next.
Please shutdown Oracle instances running out of this ORACLE_HOME on 'gr7265'.
(Oracle Home = '/u01/app/oracle/product/11.2.0/dbhome_1')
Is the node ready for patching? [y|n]
y
User Responded with: Y
Updating nodes 'gr7265'
Apply-related files are:
FP = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_files.txt"
DP = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_dirs.txt"
MP = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/make_cmds.txt"
RC = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/remote_cmds.txt"
Instantiating the file "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_files.txt.instantiated" by replacing $ORACLE_HOME in "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_files.txt" with actual path.
Propagating files to remote nodes...
Instantiating the file "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_dirs.txt.instantiated" by replacing $ORACLE_HOME in "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_dirs.txt" with actual path.
Propagating directories to remote nodes...
The node 'gr7265' has been patched. You can restart Oracle instances on it.
The node 'gr7286' will be patched next.
Please shutdown Oracle instances running out of this ORACLE_HOME on 'gr7286'.
(Oracle Home = '/u01/app/oracle/product/11.2.0/dbhome_1')
Is the node ready for patching? [y|n]
y
User Responded with: Y
Updating nodes 'gr7286'
Apply-related files are:
FP = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_files.txt"
DP = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/copy_dirs.txt"
MP = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/make_cmds.txt"
RC = "/u01/app/oracle/product/11.2.0/dbhome_1/.patch_storage/9004119_Oct_22_2009_12_29_39/rac/remote_cmds.txt"
Propagating files to remote nodes...
Propagating directories to remote nodes...
The node 'gr7286' has been patched. You can restart Oracle instances on it.
OPatch succeeded.

— The End —

WORD文档代码高亮

OoNiceDream — Thu, 25 Feb 2010 09:08:19 +0000

最近在写WORD文档，想给SQL代码高亮一下，找到以下方法，以WORD 2007为例：
样式 –> 新建样式，内容按以下填写：
名称：代码高亮
样式类型：段落
样式基准：无样式
后续段落样式：代码高亮
格式：边框–>选择一个边框，切换到底纹–>填充，选择一种颜色

OK了，来一张效果图：

跟博客的代码高亮效果差不多。

— The End —