通过RMAN进行数据库恢复（step by step）

1、修改pfile文件

使用在生产端生成的pfile文件进行修改。

PROD.__db_cache_size=1459617792

PROD.__java_pool_size=16777216

PROD.__large_pool_size=16777216

PROD.__shared_pool_size=314572800

PROD.__streams_pool_size=33554432

*._allow_level_without_connect_by=true

*._log_io_size=524288

*.audit_file_dest='/oratest/10g/PROD/admin/adump'

*.background_dump_dest='/oratest/10g/PROD/admin/bdump'

*.compatible='10.2.0.3.0'

*.control_files='/oratest/10g/PROD/control01.ctl','/oratest/10g/PROD/control02.ctl','/oratest/10g/PROD/control03.ctl'

*.core_dump_dest='/oratest/10g/PROD/admin/cdump'

*.db_block_size=8192

*.db_domain=''

*.db_file_multiblock_read_count=16

*.db_name='PROD'

*.dispatchers='(PROTOCOL=TCP) (SERVICE=PRODXDB)'

*.job_queue_processes=10

*.local_listener=''

*.lock_sga=TRUE

*.log_archive_dest_1='location=/oratest/10g/PROD/archive/'

*.log_buffer=1024000

*.open_cursors=800

*.pga_aggregate_target=524288000

*.processes=800

*.recyclebin='OFF'

*.remote_login_passwordfile='EXCLUSIVE'

*.sessions=885

*.sga_max_size=1073741824

*.sga_target=1073741824

*.shared_pool_reserved_size=62914560

*.shared_pool_size=314572800

*.undo_management='AUTO'

*.undo_retention=1800

*.undo_tablespace='UNDOTBS2'

*.user_dump_dest='/oratest/10g/PROD/admin/udump'

2、启动数据库实例，即NOMOUNT阶段

SQL> startup nomount pfile='/oratest/10g/PROD/PROD_pfile.ora';

ORACLE instance started.

Total System Global Area 1073741824 bytes

Fixed Size 2078368 bytes

Variable Size 327158112 bytes

Database Buffers 738197504 bytes

Redo Buffers 6307840 bytes

3、恢复控制文件

RMAN> restore controlfile from '/oratest/rmanbak/PROD-c-1503380933-20120728-00.ctl';

Starting restore at 28-JUL-12

using channel ORA_DISK_1

channel ORA_DISK_1: restoring control file

channel ORA_DISK_1: restore complete, elapsed time: 00:00:26

output filename=/oratest/10g/PROD/control01.ctl

output filename=/oratest/10g/PROD/control02.ctl

output filename=/oratest/10g/PROD/control03.ctl

Finished restore at 28-JUL-12

4、加载数据库，即启动到MOUNT阶段

RMAN> alter database mount;

database mounted

5、重命名redo log文件

由于恢复的数据库和原生产数据库的目录有所变动，所以需要对redo log重命名。

alter database rename file '/oradata/10g/PROD/redo01.log' to '/oratest/10g/PROD/redo01_a.log';

alter database rename file '/oradata/10g/PROD/redo02.log' to '/oratest/10g/PROD/redo02_a.log';

alter database rename file '/oradata/10g/PROD/redo03.log' to '/oratest/10g/PROD/redo03_a.log';

alter database rename file '/oradata/10g/PROD/redo04.log' to '/oratest/10g/PROD/redo04_a.log';

alter database rename file '/oradata/10g/PROD/redo05.log' to '/oratest/10g/PROD/redo05_a.log';

alter database rename file '/oradata/10g/PROD/redo6.log' to '/oratest/10g/PROD/redo06_a.log';

6、CATALOG数据库的备份片（backuppiece）

RMAN > catalog backuppiece '/oratest/rmanbak/PROD-j6nh68e9_1_1.db';

7、RESTORE数据库

由于恢复的数据文件路径与原生产端不一致，因此在RESTORE前，需要SET NEWNAME。

RMAN> run {

set newname for datafile 1 to '/oratest/10g/PROD/system01.dbf';

set newname for datafile 3 to '/oratest/10g/PROD/sysaux01.dbf';

set newname for datafile 4 to '/oratest/10g/PROD/users01.dbf';

set newname for datafile 5 to '/oratest/10g/PROD/IFSAPP_ARCHIVE_DATA.DBF';

set newname for datafile 6 to '/oratest/10g/PROD/IFSAPP_ARCHIVE_INDEX.DBF';

set newname for datafile 7 to '/oratest/10g/PROD/IFSAPP_DATA.DBF';

set newname for datafile 8 to '/oratest/10g/PROD/IFSAPP_INDEX.DBF';

set newname for datafile 9 to '/oratest/10g/PROD/IFSAPP_REPORT_DATA.DBF';

set newname for datafile 10 to '/oratest/10g/PROD/IFSAPP_REPORT_INDEX.DBF';

set newname for datafile 11 to '/oratest/10g/PROD/INDX.DBF';

set newname for datafile 12 to '/oratest/10g/PROD/PERFSTAT.DBF';

set newname for datafile 13 to '/oratest/10g/PROD/RBS.DBF';

set newname for datafile 14 to '/oratest/10g/PROD/TOOLS.DBF';

set newname for datafile 15 to '/oratest/10g/PROD/undotbs02.dbf';

restore datafile 1,3,4,5,6,7,8,9,10,11,12,13,14,15;

switch datafile all;

}

8、CATALOG ARCHIVELOG

SQL> catalog backuppiece '/oratest/rmanbak/PROD-j5nh67ta_1_1.arc';

9、RECOVER数据库

RMAN> recover database;

10、以resetlog方式打开数据库

SQL> alter database open resetlogs;