冷备手工完全恢复（recover database，recover tablespace，recover datafile）
冷备手工完全恢复
1. 手工完全恢复三种级别：
recover database： 所有或大部分datafile丢失，一般是在mount状态完成。
recover tablespace： 非关键表空间损坏，表空间下某些数据文件不能访问，一般是在open下完成。
recover datafile： 单一或少数数据文件损坏，可以在mount或open 状态完成。
四个关键文件：1）system01.dbf, 2) undo tablespace，3）control file 4）current log file
2. 手工完全恢复前提: 1）有一套datafile全备, 2）使用当前控制文件, 3）自上次备份以来的归档日志和当前联机日志是完整的
3. 实验1：（recover database ）
3.1 查看数据库当前状态，准备好冷备。
SQL> select * from andy;
ID
----------
1
--生成冷备脚本
SQL> select 'ho cp ' || name || ' /home/oracle/coldbak' from v$controlfile;
'HOCP'||NAME||'/HOME/ORACLE/COLDBAK'
--------------------------------------------------------------------------------
ho cp /home/oracle/app/oradata/orcl/control01.ctl /home/oracle/coldbak
ho cp /home/oracle/app/flash_recovery_area/orcl/control02.ctl /home/oracle/coldbak
SQL> select 'ho cp ' || name || ' /home/oracle/coldbak' from v$datafile;
'HOCP'||NAME||'/HOME/ORACLE/COLDBAK'
--------------------------------------------------------------------------------
ho cp /home/oracle/app/oradata/orcl/system01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/sysaux01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/undotbs01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/users01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/tbtb01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/ogg01.dbf /home/oracle/coldbak
6 rows selected.
SQL> shutdown immediate;
Database closed.
Database dismounted.
ORACLE instance shut down.
--冷备
SQL> ho cp /home/oracle/app/oradata/orcl/control01.ctl /home/oracle/coldbak
SQL> ho cp /home/oracle/app/flash_recovery_area/orcl/control02.ctl /home/oracle/coldbak
SQL> ho cp /home/oracle/app/oradata/orcl/system01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/sysaux01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/undotbs01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/users01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/tbtb01.dbf /home/oracle/coldbak
ho cp /home/oracle/app/oradata/orcl/ogg01.dbf /home/oracle/coldbak
--检查冷备
[oracle@11g coldbak]$ ll
total 1997776
-rw-r-----. 1 oracle oinstall 9748480 Dec 10 06:22 control01.ctl
-rw-r-----. 1 oracle oinstall 9748480 Dec 10 06:25 control02.ctl
-rw-r-----. 1 oracle oinstall 408748032 Dec 10 06:32 ogg01.dbf
-rw-r-----. 1 oracle oinstall 639639552 Dec 10 06:31 sysaux01.dbf
-rw-r-----. 1 oracle oinstall 734011392 Dec 10 06:31 system01.dbf
-rw-r-----. 1 oracle oinstall 104865792 Dec 10 06:32 tbtb01.dbf
-rw-r-----. 1 oracle oinstall 99622912 Dec 10 06:32 undotbs01.dbf
-rw-r-----. 1 oracle oinstall 39329792 Dec 10 06:32 users01.dbf
SQL> startup;
ORACLE instance started.
SQL> insert into andy values(2);
1 row created.
SQL> commit;
Commit complete.
SQL> select * from andy;
ID
----------
1
2
3.2 模拟介质失败，所有数据文件丢失
[oracle@11g orcl]$ rm -rf *.dbf //数据库在打开的情况下就删掉
SQL> shutdown immediate;
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup;
ORACLE instance started.
Total System Global Area 1068937216 bytes
Fixed Size 2220200 bytes
Variable Size 729812824 bytes
Database Buffers 331350016 bytes
Redo Buffers 5554176 bytes
Database mounted.
ORA-01157: cannot identify/lock data file 1 - see DBWR trace file
ORA-01110: data file 1: '/home/oracle/app/oradata/orcl/system01.dbf'
3.3 完全恢复流程
--查看需要恢复的文件
SQL> select file#,error from v$recover_file;
FILE# ERROR
---------- -----------------------------------------------------------------
1 FILE NOT FOUND
2 FILE NOT FOUND
3 FILE NOT FOUND
4 FILE NOT FOUND
5 FILE NOT FOUND
6 FILE NOT FOUND
6 rows selected.
SQL> select file#,checkpoint_change# from v$datafile;
FILE# CHECKPOINT_CHANGE#
---------- ------------------
1 1969481
2 1969481
3 1969481
4 1969481
5 1969481
6 1969481
6 rows selected.
SQL> select file#,checkpoint_change# from v$datafile_header;
FILE# CHECKPOINT_CHANGE#
---------- ------------------
1 0
2 0
3 0
4 0
5 0
6 0
6 rows selected.
a 首先还原所有数据文件
[oracle@11g orcl]$ cp /home/oracle/coldbak/*.dbf /home/oracle/app/oradata/orcl/
b 恢复database
SQL> recover database;
Media recovery complete.
c 打开数据库
SQL> alter database open;
Database altered.
d 验证
SQL> select * from andy;
ID
----------
1
2
实验2： recover tablespace （状态：database open)
说明：针对的是非关键表空间的损坏，基于表空间的完全恢复实际上还是对其下的datafile的恢复
模拟这种情形非常实用，通常是某个非关键表空间下的数据文件受损，但并没有造成Oracle崩溃, 我们只需针对个别有问题的tablespace去做单独的在线恢复操作，也就是说恢复时数据库整体是online的,而局部表空间是offline的，数据库不需要shutdown。
1)了解一下当前状态，有个 LZY 表空间
SQL> col file_name for a60
SQL> select file_id,file_name,tablespace_name from dba_data_files;
FILE_ID FILE_NAME TABLESPACE_NAME
---------- ------------------------------------------------------------ -------------------------
4 /home/oracle/app/oradata/orcl/users01.dbf USERS
3 /home/oracle/app/oradata/orcl/undotbs01.dbf UNDOTBS1
2 /home/oracle/app/oradata/orcl/sysaux01.dbf SYSAUX
1 /home/oracle/app/oradata/orcl/system01.dbf SYSTEM
5 /home/oracle/app/oradata/orcl/tbtb01.dbf LZY
6 /home/oracle/app/oradata/orcl/ogg01.dbf GOLDGATE
6 rows selected.
2）准备实验数据
SQL> create table andy.andydemo(id int) tablespace lzy;
Table created.
SQL> inset into andy.andydemo values(1);
SP2-0734: unknown command beginning "inset into..." - rest of line ignored.
SQL> insert into andy.andydemo values(1);
1 row created.
SQL> commit;
Commit complete.
SQL> select * from andy.andydemo;
ID
----------
1
3）模拟表空间损坏，数据库open下，直接删除表空间下的数据文件
SQL> ho rm -rf /home/oracle/app/oradata/orcl/tbtb01.dbf
4）查证该表空间上的表不可访问了
SQL> alter system flush buffer_cache; --清除data buffer
System altered.
SQL> select * from andydemo;
select * from andydemo
*
ERROR at line 1:
ORA-01116: error in opening database file 5
ORA-01110: data file 5: '/home/oracle/app/oradata/orcl/tbtb01.dbf'
ORA-27041: unable to open file
Linux-x86_64 Error: 2: No such file or directory
Additional information: 3
5）看看scn的情况
SQL> select file#,checkpoint_change# from v$datafile;
FILE# CHECKPOINT_CHANGE#
---------- ------------------
1 1969484
2 1969484
3 1969484
4 1969484
5 1969484
6 1969484
6 rows selected.
SQL> select file#,checkpoint_change# from v$datafile_header;
FILE# CHECKPOINT_CHANGE#
---------- ------------------
1 1969484
2 1969484
3 1969484
4 1969484
5 0
6 1969484
6 rows selected.
说明：
v$datafile是查看来自控制文件里记录的数据文件scn信息。
v$datafile_header是查看数据文件头中数据文件scn信息。
6）表空间offline --immediate使表空间能立即脱机，不等Oracle对任何数据文件做检查
SQL> alter tablespace lzy offline immediate;
Tablespace altered.
7）数据库open下，使用备份还原这个表空间下的所有数据文件。
SQL> ho cp /home/oracle/coldbak/tbtb01.dbf /home/oracle/app/oradata/orcl/
8）恢复tablespace
SQL> recover tablespace lzy;
Media recovery complete.
9）使表空间online
SQL> alter tablespace lzy online;
Tablespace altered. //注意：此时数据库状态一直是open的。
10) 检验
SQL> select * from andydemo;
ID
----------
1
实验3： （recover datafile，database mount或open状态）
恢复datafile, 同实验2不同的是模拟UNDO文件损坏: 因UNDO数据文件也是关键文件，所以只能在mount状态下恢复。
1） 模拟环境：
SQL> delete andy.andy; //注意：删掉了andy并提交，老值在UNDO里。
3 rows deleted.
SQL>commit；
2）在open 状态下删除datafile
SQL> ho rm -rf /home/oracle/app/oradata/orcl/undotbs01.dbf
SQL> insert into andy.andy values(4); /undo文件已经不在了，dbwr未来得及刷新，此条记录在redo里
SQL>commit；
3）关闭数据库
SQL> shutdown abort;
ORACLE instance shut down. //abort埋下伏笔，等到完全恢复时会做UNDO回滚。
4) 启动数据库mount
SQL> startup;
ORACLE instance started.
Total System Global Area 1068937216 bytes
Fixed Size 2220200 bytes
Variable Size 729812824 bytes
Database Buffers 331350016 bytes
Redo Buffers 5554176 bytes
Database mounted.
ORA-01157: cannot identify/lock data file 3 - see DBWR trace file
ORA-01110: data file 3: '/home/oracle/app/oradata/orcl/undotbs01.dbf'
5）还原并恢复UNDO数据文件
SQL> ho cp /home/oracle/coldbak/undotbs01.dbf /home/oracle/app/oradata/orcl/undotbs01.dbf
SQL> recover datafile 3
Media recovery complete.
6）打开数据库（会完成UNDO表空间数据的回滚）
SQL> alter database open;
Database altered.
7) 验证
SQL> select * from andy;
ID
----------
4
总结：有全冷备条件下三种级别恢复流程。 （相应文件名视自己的环境而定）
recover database:
SQL> select file#,error from v$recover_file;
SQL> select file#,checkpoint_change# from v$datafile;
SQL> select file#,checkpoint_change# from v$datafile_header;
SQL> ho cp /home/oracle/coldbak/*.dbf /home/oracle/app/oradata/orcl/
SQL> recover database;
SQL> alter database open;
recover tablespace:
SQL> select file#,error from v$recover_file;
SQL> select file_id,file_name,tablespace_name from dba_data_files;
SQL> select file#,checkpoint_change# from v$datafile;
SQL> select file#,checkpoint_change# from v$datafile_header;
SQL> alter tablespace lzy offline immediate;
SQL> ho cp /home/oracle/coldbak/tbtb01.dbf /home/oracle/app/oradata/orcl/
SQL> recover tablespace lzy;
SQL> alter tablespace lzy online;
recover datafile:
SQL> select file#,error from v$recover_file;
SQL> select file#,checkpoint_change# from v$datafile;
SQL> select file#,checkpoint_change# from v$datafile_header;
SQL> ho cp /home/oracle/coldbak/undotbs01.dbf /home/oracle/app/oradata/orcl/undotbs01.dbf
SQL> recover datafile 3
SQL> alter database open;
OK，转载请标明出处。
