Oracle Logminer 数据挖掘
Logminer是oracle从8i开始提供的用于分析重做日志信息的工具，它包括DBMS_LOGMNR和DBMS_LOGMNR_D两个package，后边的D是字典的意思。它既能分析redo log file，也能分析归档后的archive log file。在分析日志的过程中需要使用数据字典，一般先生成数据字典文件后使用，10g以后的版本还可以使用在线数据字典。Logminer可以分析其它数据库的重做日志文件，但是必须使用重做日志所在数据库的数据字典，否则会出现无法识别的乱码。另外被分析数据库的操作系统平台最好和当前Logminer所在数据库的运行平台一样，且block> LogMiner用于分析重做日志和归档日志所记载的事务操作。
一、确定数据库的逻辑损坏时间。假定某个用户执行drop table误删除了重要表sales，通过LogMiner可以准确定位该误操作的执行时间和SCN值，然后通过基于时间恢复或者基于SCN恢复可以完全恢复该表数据。
二、确定事务级要执行的精细逻辑恢复操作。假定某些用户在某表上执行了一系列DML操作并提交了事务，并且其中某个用户的DML操作存在错误。通过LogMiner可以取得任何用户的DML操作及相应的UNDO操作，通过执行UNDO操作可以取消用户的错误操作。
三、执行后续审计。通过LogMiner可以跟踪Oracle数据库的所有DML、DDL和DCL操作，从而取得执行这些操作的时间顺序、执行这些操作的用户等信息。
1、LogMiner基本对象
源数据库（source database）：该数据库是指包含了要分析重做日志和归档日志的产品数据库。
分析数据库（mining database）：该数据库是指执行LogMiner操作所要使用的数据库。
LogMiner字典：LogMiner字典用于将内部对象ID号和数据类型转换为对象名和外部数据格式。使用LogMiner分析重做日志和归档日志时，应该生成LogMiner字典，否则将无法读懂分析结果。
2、LogMiner配置要求
（1）源数据库和分析数据库(源数据库和分析数据库可以是同一个数据库）
源数据库和分析数据库必须运行在相同硬件平台上；
分析数据库可以是独立数据库或源数据库；
分析数据库的版本不能低于源数据库的版本；
分析数据库与源数据库必须具有相同的字符集。
（2）LogMiner字典：LogMiner字典必须在源数据库中生成。
（3）重做日志文件
当分析多个重做日志和归档日志时，它们必须是同一个源数据库的重做日志和归档日志；
当分析多个重做日志和归档日志时，它们必须具有相同的resetlogsscn；
当分析的重做日志和归档日志必须在Oracle8.0版本以上。
3、补充日志（suppplemental logging）
重做日志用于实现例程恢复和介质恢复，这些操作所需要的数据被自动记录在重做日志中。但是，重做应用可能还需要记载其他列信息到重做日志中，记录其他列的日志过程被称为补充日志。
默认情况下，Oracle数据库没有提供任何补充日志，从而导致默认情况下LogMiner无法支持以下特征：
索引簇、链行和迁移行；
直接路径插入；
摘取LogMiner字典到重做日志；
跟踪DDL；
生成键列的SQL_REDO和SQL_UNDO信息；
LONG和LOB数据类型。
因此，为了充分利用LogMiner提供的特征，必须激活补充日志。在数据库级激活补充日志的示例如下：
SQL> conn /as sysdba
已连接。
SQL>> 数据库已更改。
注意：这激活不用重启数据库，数据库联机即可。
LOGMNR 练习
1)DML 操作分析
06:02:11 SQL> conn /as sysdba
Connected.
06:02:17 SQL> show parameter utl
NAME TYPE VALUE
------------------------------------ ---------------------- ------------------------------
create_stored_outlines string
utl_file_dir string
06:02:19 SQL> conn scott/tiger
Connected.
06:04:04 SQL> select * from dept1;
DEPTNO DNAME LOC
---------- ---------------------------- --------------------------
50 Account Beijing
60 Market ShangHai
70 Study BeiJing
80 Technolygy BeiJing
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
90 Market BeiJing
80 Sales BeiJing
80 Sales ￡?￡?￡?￡?￡?￡?
11 rows selected.
06:04:18 SQL> delete from dept1 where deptno=80;
3 rows deleted.
06:04:35 SQL> select * from dept1;
DEPTNO DNAME LOC
---------- ---------------------------- --------------------------
50 Account Beijing
60 Market ShangHai
70 Study BeiJing
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
90 Market BeiJing
8 rows selected.
06:04:42 SQL> commit;
Commit complete.
06:04:55 SQL> select * from dept1;
DEPTNO DNAME LOC
---------- ---------------------------- --------------------------
50 Account Beijing
60 Market ShangHai
70 Study BeiJing
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
90 Market BeiJing
8 rows selected.
06:04:58 SQL> delete from dept1 where deptno=90;
1 row deleted.
06:05:12 SQL> commit;
Commit complete.
06:05:19 SQL>> alter system switch logfile
*
ERROR at line 1:
ORA-01031: insufficient privileges
06:05:40 SQL> conn /as sysdba
Connected.
06:05:46 SQL>>
System> 06:05:52 SQL> select * from v$log;
GROUP# THREAD#SEQUENCE# BYTES MEMBERS ARCHIV STATUS FIRST_CHANGE# FIRST_TIME
---------- ---------- ---------- ---------- ---------- ------ -------------------------------- ------------- ------------
1 1 14 52428800 2 YES ACTIVE 1748577 02-APR-11
2 1 15 52428800 2 NO CURRENT 1756744 07-APR-11
3 1 10 52428800 2 YES INACTIVE 1663012 31-MAR-11
4 1 11 52428800 2 YES INACTIVE 1673418 01-APR-11
5 1 12 52428800 2 YES INACTIVE 1705162 02-APR-11
6 1 13 52428800 2 YES INACTIVE 1727817 02-APR-11
6 rows selected.
06:06:25 SQL> select name,sequence# from v$archived_log;
NAME SEQUENCE#
-------------------------------------------------- ----------
/disk1/arch/43_1_741450701.log 43
/disk1/arch/44_1_741450701.log 44
/disk1/arch/45_1_741450701.log 45
/disk1/arch/46_1_741450701.log 46
/disk1/arch/47_1_741450701.log 47
/disk1/arch/48_1_741450701.log 48
/disk1/arch/49_1_741450701.log 49
/disk1/arch/50_1_741450701.log 50
/disk1/arch/47_1_741450701.log 47
/disk1/arch/48_1_741450701.log 48
/disk1/arch/46_1_741450701.log 46
/disk1/arch/49_1_741450701.log 49
/disk1/arch/50_1_741450701.log 50
/disk1/arch/51_1_741450701.log 51
/disk1/arch/1_1_746079267.log 1
/disk1/arch/2_1_746079267.log 2
/disk1/arch/3_1_746079267.log 3
NAME SEQUENCE#
-------------------------------------------------- ----------
/disk1/arch/4_1_746079267.log 4
/disk1/arch/5_1_746079267.log 5
/disk1/arch/6_1_746079267.log 6
/disk1/arch/7_1_746079267.log 7
/disk1/arch/8_1_746079267.log 8
/disk1/arch/9_1_746079267.log 9
/disk1/arch/10_1_746079267.log 10
/disk1/arch/11_1_746079267.log 11
/disk1/arch/12_1_746079267.log 12
/disk1/arch/13_1_746079267.log 13
/disk1/arch/14_1_746079267.log 14
28 rows selected.
06:07:01 SQL> execute dbms_logmnr.add_logfile(logfilename=>'/disk1/arch/14_1_746079267.log',options=>dbms_logmnr.new);
PL/SQL procedure successfully completed.
06:08:19 SQL> execute dbms_logmnr.start_logmnr(options=>dbms_logmnr.dict_from_online_catalog);
PL/SQL procedure successfully completed.
06:09:08 SQL> col username for a8
06:09:26 SQL> col sql_redo for a50
06:09:36 SQL>>
Session> 06:10:03 SQL> select username,scn,timestamp,sql_redo from v$logmnr_contents where seg_name='SCOTT.DEPT1';
no rows selected
06:14:38 SQL> select username,scn,timestamp,sql_redo from v$logmnr_contents where seg_name='DEPT1';
USERNAME SCN TIMESTAMP SQL_REDO
-------- ---------- ------------ --------------------------------------------------
1756706 07-APR-11 delete from "SCOTT"."DEPT1" where "DEPTNO" = '80'
and "DNAME" = 'Technolygy' and "LOC" = 'BeiJing' a
nd ROWID = 'AAANaPAAEAAAAAcAAD';
1756707 07-APR-11 delete from "SCOTT"."DEPT1" where "DEPTNO" = '80'
and "DNAME" = 'Sales' and "LOC" = 'BeiJing' and RO
WID = 'AAANaPAAEAAAAAgAAC';
1756707 07-APR-11 delete from "SCOTT"."DEPT1" where "DEPTNO" = '80'
and "DNAME" = 'Sales' and "LOC" = '￡?￡?￡?￡?￡?￡?' a
nd ROWID = 'AAANaPAAEAAAAAgAAD';
1756728 07-APR-11 delete from "SCOTT"."DEPT1" where "DEPTNO" = '90'
and "DNAME" = 'Market' and "LOC" = 'BeiJing' and R
OWID = 'AAANaPAAEAAAAAgAAA';
06:15:09 SQL>
结束日志分析
09:18:14 SQL>execute dbms_logmnr.end_logmnr;
PL/SQL procedure successfully completed.
2）DDL 操作分析
06:20:46 SQL> connscott/tiger
Connected.
06:20:54 SQL> select * from tab;
TNAME TABTYPE CLUSTERID
------------------------------------------------------------ -------------- ----------
EMP TABLE
DEPT TABLE
BONUS TABLE
SALGRADE TABLE
QUEST_SL_TEMP_EXPLAIN1 TABLE
EMP1 TABLE
ERRLOG TABLE
PART_SALES TABLE
T01 TABLE
DEPT1 TABLE
TT01 TABLE
USERLOG TABLE
ERRORS TABLE
EMP_HIS TABLE
14 rows selected.
06:21:02 SQL> drop table emp_his purge;
Table dropped.
06:21:21 SQL> drop table tt01 purge;
Table dropped.
06:21:32 SQL> drop table t01 purge;
Table dropped.
06:21:38 SQL> select * from tab;
TNAME TABTYPE CLUSTERID
------------------------------------------------------------ -------------- ----------
EMP TABLE
DEPT TABLE
BONUS TABLE
SALGRADE TABLE
QUEST_SL_TEMP_EXPLAIN1 TABLE
EMP1 TABLE
ERRLOG TABLE
PART_SALES TABLE
DEPT1 TABLE
USERLOG TABLE
ERRORS TABLE
11 rows selected.
06:22:17 SQL> conn /as sysdba
Connected.
06:22:21 SQL>>
System> 06:22:23 SQL> show parameter utl
NAME TYPE VALUE
------------------------------------ ---------------------- ------------------------------
create_stored_outlines string
utl_file_dir string
06:22:33 SQL>>
System> shutdown immediate
Database closed.
Database dismounted.
ORACLE instance shut down.
06:23:59 SQL> startup
ORACLE instance started.
Total System Global Area524288000 bytes
Fixed>
Variable> Database Buffers 331350016 bytes
Redo Buffers 2973696 bytes
Database mounted.
Database opened.
06:24:24 SQL> show parameter utl
NAME TYPE VALUE
------------------------------------ ---------------------- ------------------------------
create_stored_outlines string
utl_file_dir string /home/oracle/logmnr
06:26:28 SQL> execute dbms_logmnr_d.build('dict.ora','/home/oracle/logmnr',dbms_logmnr_d.store_in_flat_file);
PL/SQL procedure successfully completed.
06:27:46 SQL> select name,sequence# from v$archived_log;
NAME SEQUENCE#
-------------------------------------------------- ----------
/disk1/arch/44_1_741450701.log 44
/disk1/arch/45_1_741450701.log 45
/disk1/arch/46_1_741450701.log 46
/disk1/arch/47_1_741450701.log 47
/disk1/arch/48_1_741450701.log 48
/disk1/arch/49_1_741450701.log 49
/disk1/arch/50_1_741450701.log 50
/disk1/arch/47_1_741450701.log 47
/disk1/arch/48_1_741450701.log 48
/disk1/arch/46_1_741450701.log 46
/disk1/arch/49_1_741450701.log 49
/disk1/arch/50_1_741450701.log 50
/disk1/arch/51_1_741450701.log 51
/disk1/arch/1_1_746079267.log 1
/disk1/arch/2_1_746079267.log 2
/disk1/arch/3_1_746079267.log 3
/disk1/arch/4_1_746079267.log 4
NAME SEQUENCE#
-------------------------------------------------- ----------
/disk1/arch/5_1_746079267.log 5
/disk1/arch/6_1_746079267.log 6
/disk1/arch/7_1_746079267.log 7
/disk1/arch/8_1_746079267.log 8
/disk1/arch/9_1_746079267.log 9
/disk1/arch/10_1_746079267.log 10
/disk1/arch/11_1_746079267.log 11
/disk1/arch/12_1_746079267.log 12
/disk1/arch/13_1_746079267.log 13
/disk1/arch/14_1_746079267.log 14
/disk1/arch/15_1_746079267.log 15
28 rows selected.
06:28:09 SQL>execute dbms_logmnr.add_logfile(logfilename=>'/disk1/arch/15_1_746079267.log',options=>dbms_logmnr.new);
PL/SQL procedure successfully completed.
06:28:58 SQL> execute dbms_logmnr.start_logmnr(dictfilename=>'/home/oracle/logmnr/dict.ora',options=>dbms_logmnr.ddl_dict_tracking);
PL/SQL procedure successfully completed.
06:30:58 SQL> select username,scn,timestamp,sql_redo from v$logmnr_contents
06:31:24 2 where lower(sql_redo) like '%table%';
USERNAME SCN TIMESTAMP SQL_REDO
-------- ---------- ------------ --------------------------------------------------
1757390 07-APR-11 drop table emp_his purge;
1757409 07-APR-11 drop table tt01 purge;
1757425 07-APR-11 drop table t01 purge;
06:31:45 SQL>
结束日志分析
SQL> execute dbms_logmnr.end_logmnr;
PL/SQL procedure successfully completed.
oracle视频教程请关注:http://down.51cto.com/4202939/up
页:
[1]