AG阅读总结9.1——日志管理

一、Oracle中的几类日志文件：

    Redo log files      -->联机重做日志

    Archive log files   -->归档日志

    Alert log files     -->告警日志

    Trace files         -->跟踪日志

    user_dump_dest          -->用户跟踪日志

    backupground_dump_dest  -->进程跟踪日志

   

    --查看后台进程相关目录

        SQL> show parameter dump

 

        NAME                                 TYPE        VALUE

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

        background_core_dump                 string      partial

        background_dump_dest                 string      /u01/app/oracle/admin/orcl/bdump

        core_dump_dest                       string      /u01/app/oracle/admin/orcl/cdump

        max_dump_file_size                   string      UNLIMITED

        shadow_core_dump                     string      partial

        user_dump_dest                       string      /u01/app/oracle/admin/orcl/udump  

       

    关于Oracle 常用目录及路径请参考：Oracle 常用目录结构(10g)

    关于Oracle 体系结构请参考：Oracle实例和Oracle数据库(Oracle体系结构)

二、联机重做日志的规划管理：

    1.联机重做日志     

        记录了数据的所有变化(DML,DDL或管理员对数据所作的结构性更改等)

        提供恢复机制(对于意外删除或宕机利用日志文件实现数据恢复)

        可以被分组管理

       

    2.联机重做日志组

        由一个或多个相同的联机日志文件组成一个联机重做日志组

        至少两个日志组，每组一个成员(建议每组两个成员，分散放开到不同的磁盘)

        由LGWR后台进程同时将日志内容写入到一个组的所有成员

            LGWR的触发条件：

                在事务提交的时候(COMMIT)

                Redo Log Buffer 三分之一满

                Redo Log Buffer 多于一兆的变化记录

                在DBWn写入数据文件之前

 3.联机重做日志成员

        重做日志组内的每一个联机日志文件称为一个成员

        一个组内的每一个成员具有相同的日志序列号(log sequence number),且成员的大小相同

        每次日志切换时，Oracle服务器分配一个新的LSN号给即将写入日志的日志文件组

        LSN号用于唯一区分每一个联机日志组和归档日志

        处于归档模式的联机日志，LSN号在归档时也被写入到归档日志之中

       

    4.日志文件的工作方式

        日志文件采用按顺序循环写的方式

        当一组联机日志组写满，LGWR则将日志写入到下一组，当最后一组写满则从第一组开始写入

        写入下一组的过程称为日志切换

        切换时发生检查点过程

        检查点的信息同时写入到控制文件

   

    5.联机日志文件的规划

        总原则

            分散放开，多路复用

            日志所在的磁盘应当具有较高的I/O

            一般日志组大小应满足自动切换间隔至少15-20分钟左右业务需求

            建议使用rdo结尾的日志文件名，避免误删日志文件。如redo1.rdo,redo2.rdo

        规划样例

            Redo Log Group1     Redo Log Group2     Redo Log Group3

           

            Member1             Member1              Member1            -->Physical Disk 1

           

            Member2             Member2              Member2            -->Physical Disk 2

           

            Member3             Member3              Member3            -->Physical Disk 3

           

    6.日志切换和检查点切换

        ALTER SYSTEM SWITCH LOGFILE;   --强制手动切换

       

        ALTER SYSTEM CHECKPOINT;

       

        强制设置检查点间隔

        ALTER SYSTEM SET FAST_START_MTTR_TARGET = n

       

    7.添加日志文件组

        ALTER DATABASE ADD LOGFILE [GROUP n]

            ('$ORACLE_BASE/oradata/u01/logn1.rdo',

             '$ORACLE_BASE/oradata/u01/logn2.rdo')

            SIZE mM;

alter database add logfile group 3('/u01/prod/db/apps_st/data/log03a.dbf','/u01/prod/db/apps_st/data/log03b.dbf')SIZE 1024M;

           

    8.添加日志成员

        ALTER DATABASE ADD LOGFILE MEMBER

        '$ORACLE_BASE/oradata/u01/logn1.rdo' TO GROUP 1,

        '$ORACLE_BASE/oradata/u01/logn2.rdo' TO GROUP 2;

       

    9.删除日志成员

        不能删除组内的唯一一个成员

        不能删除处于active 和current 状态组内的成员

        删除处于active 和current 状态组内的成员，应使用日志切换使其处于INACTIVE状态后再删除

        对于组内如果一个成员为NULL 值，一个为INVALID，且组处入INACTIVE，仅能删除INVALID状态成员

        删除日志成员，物理文件并没有真正删除，需要手动删除

        删除日志文件后，控制文件被更新

        对于处于归档模式下的数据库，删除成员时确保日志已被归档，查看v$log视图获得归档信息

        ALTER DATABASE DROP LOGFILE MEMBER '$ORACLE_BASE/oradata/u01/logn1.rdo'

       

    10.删除日志组

        一个实例至少需要两个联机日志文件组

        活动或当前的日志组不能被删除

        组内成员状态有NULL 值或INVALID状态并存，组不可删除

        日志组被删除后，物理文件需要手动删除(对于非OMF)

        ALTER DATABASE DROP LOGFILE GROUP n

       

    11.日志的重定位及重命名

        所需权限

            ALTER DATABASE 系统权限

            复制文件到目的位置操作系统权限(写权限)

        CURRENT状态组内的成员不能被重命名

        建议该行为之前备份数据库

        重命名或重定位之后建议立即备份控制文件

        重定位及重命名的两种方法

            添加一个新成员到日志组，然后删除一个旧的成员

            使用ALTER DATABASE RENAME FILE 命令(不区分归档与非归档模式)

                复制联机日志文件到新路径:ho cp

                执行ALTER DATABASE RENAME FILE '' TO ''

            对于处于CURRENT状态的需要改名且不切换的情况下

                办法是切换到MOUNT状态下再执行上述操作

                               

    12.清空日志文件组

        ALTER DATABASE CLEAR LOGIFLE GROUP n

        ALTER DATABASE CLEAR UNARCHIVED LOGFILE GROUP n --使用unarchived 避免归档

   

    13.日志周期循环及切换分析

       

        Group 1      Group 2      Group 3

       

        Current      Inactive     Inactive

        ---------- Log Switch -------------

        Active       Current      Inactive

        ---------- Log Switch -------------

        Active       Active       Current

        ---------- Log Switch -------------

        Current      Inactive     Inactive

       

        --Active 和Current 称之为在一个循环周期之内(按顺序写日志)

        --Inactive 称为一个周期之外(一个新的循环)

        --新一轮循环开始如在归档状态则先归档再清空，否则直接清空日志

        --数据库启动时Active 和Current 状态的日志不能丢失，否则出错

       

    14.日志的监视

        查看日志视图中的物理日志文件是否存在、位置、大小等

            SELECT 'ho cp '||member FROM v$logfile;

        查看日志文件所处的磁盘空间是否足够

            SQL> ho df -h

        查看组内是否存在多个成员，如为单一成员应考虑增加日志成员

        日志切换的间隔时间，应满足15-20分钟业务需求，如果切换间隔很短，应当增加日志文件的大小

            增加方法，先删除日志组，再重建该组(对于current和active的需要切换再做处理)

 

        --查看切换时间间隔(下面的示例中为手工切换的时间，不作考虑)

        SQL> SELECT TO_CHAR(first_time,'yyyy-mm-dd hh24:mi:ss'),group# FROM v$log;

   

        TO_CHAR(FIRST_TIME,     GROUP#

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

        2010-07-20 09:43:18          1

        2010-07-19 22:44:30          2

        2010-07-19 22:44:32          3         

       

    15.日志的异常处理(参照演示中9小节)

        不一致的情况(启动时)

            ALTER DATABASE CLEAR LOGFILE GROUP n

            ALTER DATABASE CLEAR UNARCHIVED LOGFILE GROUP n

三、与日志有关的动态性能视图

        V$LOG

        V$LOGFILE

       

        V$LOG中STATUS的状态值

            UNUSED: 从未对该联机日志写入任何内容，一般为新增加联机日志文件或是使用resetlog后的状态

            CURRENT：当前重做日志文件，表示该重做日志文件为活动状态，能够被打开和关闭

            ACTIVE：处于活动状态，不属于当前日志，崩溃恢复需要该状态，可用于块恢复，可能归档，也可能未归档

            CLEARING：表示在执行alter database clear logfile命令后正将该日志重建为一个空日志，重建后状态变为unused

            CLEARING_CURRENT:当前日志处于关闭线程的清除状态。如日志某些故障或写入新日志标头时发生I/O错误

            INACTIVE:实例恢复不在需要联机重做文件日志组，可能归档也可能未归档

           

        V$LOGFILE中STATUS的状态值

            INVALID :表明该文件不可访问

            STALE :表明文件内容不完全

            DELETED : 表明该文件不再使用

            NULL ：表明文件正在使用

 

四、演示

    --1.查看当前数据库的日志

        SQL> SELECT * FROM v$log;

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          5   52428800          1 NO  CURRENT                2758062 19-JUL-10

                 2          1          3   52428800          2 YES INACTIVE               2695010 16-JUL-10

                 3          1          4  104857600          2 YES INACTIVE               2716552 18-JUL-10

                 

 

        SQL> SELECT * FROM v$logfile ORDER BY group#;  

 

            GROUP# STATUS  TYPE    MEMBER                                                  IS_

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

                 1         ONLINE  /u01/app/oracle/oradata/orcl/redo01.log                 NO

                 2 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo02.log                 NO

                 2 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo2.log                  NO

                 3 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo03.log                 NO

                 3 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo3.log                  NO  

                 

    --2.添加日志组

        SQL> SELECT * FROM v$logfile;

 

            GROUP# STATUS  TYPE    MEMBER                                                  IS_

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

                 2 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo2.log                  NO

                 2 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo02.log                 NO

                 1         ONLINE  /u01/app/oracle/oradata/orcl/redo01.log                 NO

                 3 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo3.log                  NO

                 3 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo03.log                 NO

                 4         ONLINE  /u01/app/oracle/oradata/orcl/redo4.log                  NO

                 4         ONLINE  /u01/app/oracle/oradata/orcl/redo04.log                 NO

自动 增加组名：

          SQL> ALTER DATABASE  ADD LOGFILE ('/oracle/dbs/log1c.rdo', '/oracle/dbs/log2c.rdo') SIZE 500K;

指定组名：

          SQL> ALTER DATABASE ADD LOGFILE GROUP 10 ('/oracle/dbs/log1c.rdo', '/oracle/dbs/log2c.rdo')
SIZE 8192k;     

    --3.添加日志成员

        SQL> ALTER DATABASE ADD LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo1.log' TO GROUP 1;

 

        Database altered.

      --用命名规则直接加到某个组中，我们添加是log2c.rdo,原有的是log2a.rdo和log2b.rdo.所以会自动添加到这两个对应的组中。

        SQL> ALTER DATABASE ADD LOGFILE MEMBER '/oracle/dbs/log2c.rdo' TO ('/oracle/dbs/log2a.rdo', '/oracle/dbs/log2b.rdo');

 

        SQL> SELECT * FROM v$logfile WHERE group# = 1 ;

 

            GROUP# STATUS  TYPE    MEMBER                                                  IS_

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

                 1         ONLINE  /u01/app/oracle/oradata/orcl/redo01.log                 NO

                 1 INVALID ONLINE  /u01/app/oracle/oradata/orcl/redo1.log                  NO

                 

    --4.删除日志成员

        SQL> ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo01.log';

        ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo01.log'

        *

        ERROR at line 1:  --redo01.log处于NULL状态且该日志组为current状态不能删除

        ORA-00362: member is required to form a valid logfile in group 1

        ORA-01517: log member: '/u01/app/oracle/oradata/orcl/redo01.log'

 

 

        SQL> ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo04.log';

 

        Database altered.

 

        SQL>  ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo4.log';

         ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo4.log'

        *

        ERROR at line 1:   --最后一个日志成员不能被删除

        ORA-00361: cannot remove last log member /u01/app/oracle/oradata/orcl/redo4.log for group 4

 

    --5.日志切换

        SQL> SELECT * FROM v$log;  --当前的日志组处于CURRENT状态

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          5   52428800          2 NO  CURRENT                2758062 19-JUL-10

                 2          1          3   52428800          2 YES INACTIVE               2695010 16-JUL-10

                 3          1          4  104857600          2 YES INACTIVE               2716552 18-JUL-10

                 4          1          0   31457280          1 YES UNUSED                       0

 

        SQL> ALTER SYSTEM SWITCH LOGFILE;  --进行日志切换

 

        System altered.

 

        SQL> SELECT * FROM v$log;    --原来的日志组4的unused状态变为current状态

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          5   52428800          2 YES ACTIVE                 2758062 19-JUL-10

                 2          1          3   52428800          2 YES INACTIVE               2695010 16-JUL-10

                 3          1          4  104857600          2 YES INACTIVE               2716552 18-JUL-10

                 4          1          6   31457280          1 NO  CURRENT                2759277 19-JUL-10

 

        SQL> ALTER SYSTEM SWITCH LOGFILE;   --再次进行日志切换

 

        System altered.

 

        SQL> SELECT * FROM v$log;          --日志组1变为current且组4变为active 状态

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          5   52428800          2 YES ACTIVE                 2758062 19-JUL-10

                 2          1          7   52428800          2 NO  CURRENT                2759293 19-JUL-10

                 3          1          4  104857600          2 YES INACTIVE               2716552 18-JUL-10

                 4          1          6   31457280          1 YES ACTIVE                 2759277 19-JUL-10       

 

    由上可得知，在日志切换时对于unused组将优先作为下一组切换对象

        --再次删除redo01.log还是收到错误提示

        SQL> ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo01.log';

        ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo01.log'

        *

        ERROR at line 1:

        ORA-00362: member is required to form a valid logfile in group 1

        ORA-01517: log member: '/u01/app/oracle/oradata/orcl/redo01.log'

 

        SQL> ALTER SYSTEM SWITCH LOGFILE;  --再次进行日志切换

 

        System altered.

 

        SQL> SELECT * FROM v$log;          --group1变为inactive

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          5   52428800          2 YES INACTIVE               2758062 19-JUL-10

                 2          1          7   52428800          2 YES ACTIVE                 2759293 19-JUL-10

                 3          1          8  104857600          2 NO  CURRENT                2759420 19-JUL-10

                 4          1          6   31457280          1 YES INACTIVE               2759277 19-JUL-10

 

        --反复多切几次日志之后redo01.log被成功删除     

        SQL>  ALTER DATABASE DROP LOGFILE MEMBER '/u01/app/oracle/oradata/orcl/redo01.log';

 

    Database altered.

         

    --6.删除日志组(CURRENT和ACTIVE状态的不能被删除)

        SQL> SELECT * FROM v$log;

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          9   52428800          1 YES ACTIVE                 2759487 19-JUL-10

                 2          1         11   52428800          2 NO  CURRENT                2759502 19-JUL-10

                 3          1          8  104857600          2 YES ACTIVE                 2759420 19-JUL-10

                 4          1         10   31457280          1 YES ACTIVE                 2759499 19-JUL-10

                 

        SQL> ALTER DATABASE DROP LOGFILE GROUP 4;

        ALTER DATABASE DROP LOGFILE GROUP 4

        *

        ERROR at line 1:   --处于活动状态的group4 用于灾难恢复，不能被删除

        ORA-01624: log 4 needed for crash recovery of instance orcl (thread 1)

        ORA-00312: online log 4 thread 1: '/u01/app/oracle/oradata/orcl/redo4.log'

 

        SQL> ALTER SYSTEM SWITCH LOGFILE;   --进行日志切换

 

        System altered.

 

        SQL> /

 

        System altered.

 

        SQL> SELECT * FROM v$log;    --group 4的状态变为inactvie

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1         13   52428800          1 NO  CURRENT                2759720 19-JUL-10

                 2          1         11   52428800          2 YES ACTIVE                 2759502 19-JUL-10

                 3          1         12  104857600          2 YES ACTIVE                 2759718 19-JUL-10

                 4          1         10   31457280          1 YES INACTIVE               2759499 19-JUL-10

 

        SQL> ALTER DATABASE DROP LOGFILE GROUP 4;   --成功删除group 4

 

        Database altered.

 

        SQL> ho ls /u01/app/oracle/oradata/orcl/redo*

        /u01/app/oracle/oradata/orcl/redo01.log  /u01/app/oracle/oradata/orcl/redo1.log

        /u01/app/oracle/oradata/orcl/redo02.log  /u01/app/oracle/oradata/orcl/redo2.log

        /u01/app/oracle/oradata/orcl/redo03.log  /u01/app/oracle/oradata/orcl/redo3.log

        /u01/app/oracle/oradata/orcl/redo04.log  /u01/app/oracle/oradata/orcl/redo4.log

 

        SQL> ho rm /u01/app/oracle/oradata/orcl/redo04.log   --删除物理文件

 

        SQL> ho rm /u01/app/oracle/oradata/orcl/redo4.log    --删除物理文件

 

    --7.日志的重定位及重命名(仅演示ALTER DATABASE RENAME FILE 命令)

       官方文档重定位步骤：

1、关闭数据库：shutdown。

2、将要移动位置的文件移动到指定位置：

mv /diska/logs/log1a.rdo /diskc/logs/log1c.rdo
mv /diska/logs/log2a.rdo /diskc/logs/log2c.rdo

3、开启数据库，挂载，不open：

CONNECT / as SYSDBA
STARTUP MOUNT

4、在sqlplus中重命名新位置的redo log文件：

alter database rename file '/diska/logs/log1a.rdo' , '/diska/logs/log1b.rdo'  TO '/diskc/logs/log1c.rdo' , '/diskc/logs/log2c.rdo';

5、正常打开数据库：ALTER DATABASE OPEN.

 SQL> SELECT name,log_mode FROM v$database;

 

        NAME      LOG_MODE

        --------- ------------

        ORCL      ARCHIVELOG

 

        SQL> SELECT * FROM v$logfile ORDER BY group#;

 

            GROUP# STATUS  TYPE    MEMBER                                                  IS_

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

                 1         ONLINE  /u01/app/oracle/oradata/orcl/redo01.log                 NO

                 2         ONLINE  /u01/app/oracle/oradata/orcl/redo02.log                 NO

                 2         ONLINE  /u01/app/oracle/oradata/orcl/redo2.log                  NO

                 3 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo03.log                 NO

                 3 STALE   ONLINE  /u01/app/oracle/oradata/orcl/redo3.log                  NO

 

        SQL> ho cp /u01/app/oracle/oradata/orcl/redo01.log /u01/app/oracle/oradata/redo01.rdo      

 

        SQL> ALTER DATABASE RENAME FILE '/u01/app/oracle/oradata/orcl/redo01.log'

          2    TO '/u01/app/oracle/oradata/redo01.rdo';

 

        Database altered.

 

        SQL> SELECT * FROM v$logfile WHERE group# = 1;

 

            GROUP# STATUS  TYPE    MEMBER                                                  IS_

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

                 1         ONLINE  /u01/app/oracle/oradata/redo01.rdo                      NO

       

 

    --8.清空日志文件组(只有非active 和非current状态的组才能被清空)

        SQL> SELECT * FROM v$log;

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1         13   52428800          1 YES ACTIVE                 2759720 19-JUL-10

                 2          1         14   52428800          2 NO  CURRENT                2761383 19-JUL-10

                 3          1         12  104857600          2 YES INACTIVE               2759718 19-JUL-10

                 

        SQL> ALTER DATABASE CLEAR LOGFILE GROUP 1;

        ALTER DATABASE CLEAR LOGFILE GROUP 1

        *

        ERROR at line 1:     --active 状态不能被清空

        ORA-01624: log 1 needed for crash recovery of instance orcl (thread 1)

        ORA-00312: online log 1 thread 1: '/u01/app/oracle/oradata/redo1.rdo'

 

        SQL>  ALTER DATABASE CLEAR LOGFILE GROUP 2;

         ALTER DATABASE CLEAR LOGFILE GROUP 2

        *

        ERROR at line 1:    --current 状态不能被清空

        ORA-01624: log 2 needed for crash recovery of instance orcl (thread 1)

        ORA-00312: online log 2 thread 1: '/u01/app/oracle/oradata/orcl/redo2.log'

        ORA-00312: online log 2 thread 1: '/u01/app/oracle/oradata/orcl/redo02.log'

 

 

        SQL>  ALTER DATABASE CLEAR LOGFILE GROUP 3;

 

        Database altered.

 

 

        SQL> SELECT * FROM v$log;   --group 3被清空后状态变为unused

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1         13   52428800          1 YES INACTIVE               2759720 19-JUL-10

                 2          1         14   52428800          2 NO  CURRENT                2761383 19-JUL-10

                 3          1          0  104857600          2 YES UNUSED                 2759718 19-JUL-10

             

    --9.日志异常处理

        --启动时提示日志不一致

        SQL> startup

        ORACLE instance started.

 

        Total System Global Area  251658240 bytes

        Fixed Size                  1218796 bytes

        Variable Size              83887892 bytes

        Database Buffers          163577856 bytes

        Redo Buffers                2973696 bytes

        Database mounted.

        ORA-00341:log 1 of thread 1,wrong log # in header

        ORA-00312:online log 1 thread 1:'/u01/app/oracle/oradata/orcl/redo1a.rdo'  

        ORA-00312:online log 1 thread 1:'/u01/app/oracle/oradata/orcl/redo1b.rdo'

 

        SQL> ALTER DATABASE CLEAR LOGFILE GROUP 1;

 

        Database altered.

 

        SQL> ALTER DATABASE OPEN;

 

        Database opened.

 

        --日志文件丢失(非current状态日志组)

        SQL> startup

        ORACLE instance started.

 

        Total System Global Area  251658240 bytes

        Fixed Size                  1218796 bytes

        Variable Size              88082196 bytes

        Database Buffers          159383552 bytes

        Redo Buffers                2973696 bytes

        Database mounted.

        ORA-00313: open failed for members of log group 1 of thread 1

        ORA-00312: online log 1 thread 1: '/u01/app/oracle/oradata/orcl/redo1a.rdo'

        ORA-00312: online log 1 thread 1: '/u01/app/oracle/oradata/orcl/redo1b.rdo'

 

        SQL> ALTER DATABASE CLEAR LOGFILE GROUP 1;

 

        Database altered.

 

        SQL> ALTER DATABASE OPEN;

 

        Database altered.

 

        --日志文件丢失(current状态日志组)

        SQL> startup

        ORACLE instance started.

 

        Total System Global Area  251658240 bytes

        Fixed Size                  1218796 bytes

        Variable Size              83887892 bytes

        Database Buffers          163577856 bytes

        Redo Buffers                2973696 bytes

        Database mounted.

        ORA-00313: open failed for members of log group 3 of thread 1

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3a.rdo'

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3b.rdo'

 

        --查看告警日志

        SQL> ho tail -n 30 /u01/app/oracle/admin/orcl/bdump/alert_orcl.log

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3a.rdo'

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

        Tue Jul 20 10:45:58 2010

        Errors in file /u01/app/oracle/admin/orcl/bdump/orcl_lgwr_4112.trc:

        ORA-00313: open failed for members of log group 3 of thread 1

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3b.rdo'

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3a.rdo'

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

        Tue Jul 20 10:45:58 2010

        ARC0: STARTING ARCH PROCESSES

        Tue Jul 20 10:45:58 2010

        ARC1: Becoming the 'no FAL' ARCH

        ARC1: Becoming the 'no SRL' ARCH

        Tue Jul 20 10:45:58 2010

        ARC2: Archival started

        ARC0: STARTING ARCH PROCESSES COMPLETE

        ARC0: Becoming the heartbeat ARCH

        ARC2 started with pid=18, OS id=4137

        Tue Jul 20 10:45:58 2010

        ORA-313 signalled during: ALTER DATABASE OPEN...

       

        --查看物理日志文件是否存在

        SQL> ho ls /u01/app/oracle/oradata/orcl/redo3a.rdo

        ls: /u01/app/oracle/oradata/orcl/redo3a.rdo: No such file or directory

 

        SQL> ho ls /u01/app/oracle/oradata/orcl/redo3b.rdo

        ls: /u01/app/oracle/oradata/orcl/redo3b.rdo: No such file or directory

       

--尝试使用清空日志组命令

        SQL> ALTER DATABASE CLEAR LOGFILE GROUP 3;

        ALTER DATABASE CLEAR LOGFILE GROUP 3

        *

        ERROR at line 1:   --系统处于非归档模式，且group 3状态为CURRENT

        ORA-00350: log 3 of instance orcl (thread 1) needs to be archived

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3a.rdo'

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3b.rdo'

 

        --尝试使用不归档清空日志

        SQL> ALTER DATABASE CLEAR UNARCHIVED LOGFILE GROUP 3;

        ALTER DATABASE CLEAR UNARCHIVED LOGFILE GROUP 3

        *

        ERROR at line 1:  

        ORA-00313: open failed for members of log group 3 of thread 1

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3b.rdo'

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

        ORA-00312: online log 3 thread 1: '/u01/app/oracle/oradata/orcl/redo3a.rdo'

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

 

        --使用带控制文件的介质恢复

        SQL> RECOVER DATABASE USING BACKUP CONTROLFILE;

        ORA-00279: change 2835232 generated at 07/20/2010 10:40:23 needed for thread 1

        ORA-00289: suggestion : /u01/app/oracle/flash_recovery_area/ORCL/archivelog/2010_07_20/o1_mf_1_39_%u_.arc

        ORA-00280: change 2835232 for thread 1 is in sequence #39

 

 

        Specify log: {<RET>=suggested | filename | AUTO | CANCEL}

 

        ORA-00308: cannot open archived log '/u01/app/oracle/flash_recovery_area/ORCL/archivelog/2010_07_20/o1_mf_1_39_%u_.arc'

        ORA-27037: unable to obtain file status

        Linux Error: 2: No such file or directory

        Additional information: 3

 

        --使用resetlogs选项打开数据库

        SQL> ALTER DATABASE OPEN RESETLOGS;

 

        Database altered.      

 

        SQL> SELECT * FROM v$log;   --系统重建group 3

 

            GROUP#    THREAD#  SEQUENCE#      BYTES    MEMBERS ARC STATUS           FIRST_CHANGE# FIRST_TIM

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

                 1          1          2   31457280          2 NO  CURRENT                2835234 20-JUL-10

                 2          1          1   31457280          2 YES INACTIVE               2835233 20-JUL-10

                 3          1          0   31457280          2 YES UNUSED                       0          

                 

        SQL> SELECT * FROM v$logfile;   --为group 3增加了两个成员redo3a.rdo ，redo3b.rdo  

 

            GROUP# STATUS  TYPE    MEMBER                                                  IS_

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

                 2         ONLINE  /u01/app/oracle/oradata/orcl/redo2a.rdo                 NO

                 2         ONLINE  /u01/app/oracle/oradata/orcl/redo2b.rdo                 NO

                 1         ONLINE  /u01/app/oracle/oradata/orcl/redo1a.rdo                 NO

                 3         ONLINE  /u01/app/oracle/oradata/orcl/redo3a.rdo                 NO

                 3         ONLINE  /u01/app/oracle/oradata/orcl/redo3b.rdo                 NO

                 1         ONLINE  /u01/app/oracle/oradata/orcl/redo1b.rdo                 NO      

 

        对于CURRENT组的也可以使用隐藏参数来解决

        步骤：

            alter system set "_allow_resetlogs_corruption" = true scope = spfile;

           

            recover database using bakcup controlfile;

           

            alter database open resetlogs;

           

            shutdown immediate;

           

            startup mount;

           

            alter database open resetlogs;

           

            alter system reset "_allow_resetlogs_corruption" scope = spfile sid = '*'

   

        对于归档模式下的日志文件丢失，同样可以按上述步骤处理