如何实施一次完整的oracle异机恢复

由于以后工作可能会有变化，为了防止自己记忆力衰退
将oracle通过nbu做的异机整库恢复记录下来

首先要保证操作系统平台大版本一致（从HP 11.31到11.11进行恢复也没有问题），如果是跨平台的恢复这里不做讨论
oracle版本必须一致，包括小版本号

假设恢复的源主机hostname叫 hosta，目标主机hostname叫 hostb，数据库sid叫 orasid

先保证NBU的设置允许异机恢复
----------------------------------------------------------------------
确保将 NetBackup 服务器配置为允许重定向恢复。管理员通过在 Netbackup 主服务器上创建下面的文件可删除对所客户机的限制：
/usr/openv/netbackup/db/altnames/No.Restrictions
或者，要限制客户机只从其他特定的客户机恢复，请创建下面的文件：
/usr/openv/netbackup/db/altnames/client_name
其中 client_name 是允许进行重定向恢复的客户机（目标客户机）的名称。然后，将NetBackup for Oracle 源客户机的名称添加到此文件。

本节信息适用于使用命令行执行的恢复，而不适用于“备份、归档和恢复”客户机界面
----------------------------------------------------------------------

1、在NBU Master上用bplist命令查出已经备份的信息
bplist -C hosta -t 4 -R |more
。。。

记录下对应的controlfile备份信息，以后备用，这里是c-297293101-20110427-01

2、配置好hostb的oracle，安装完成，并且按照hosta的目录结构建好目录和文件系统

3、配置好hostb的nbu，完成oracle_link

接下来的4到6步是进行spfile的恢复动作

4、启动hostb上的oracle到nomount状态，这时使用的是安装完成默认的pfile文件

[oracle@hostb:/u01/oracle/product/db10gr2/dbs]$sqlplus /nolog

SQL*Plus: Release 10.2.0.4.0 - Production on Wed Apr 27 15:58:31 2011

Copyright (c) 1982, 2007, Oracle.  All Rights Reserved.

SQL> connect /as sysdba;
Connected to an idle instance.
SQL> startup nomount pfile='/u01/oracle/product/db10gr2/dbs/init.ora';
ORACLE instance started.

Total System Global Area  171966464 bytes
Fixed Size                  2067664 bytes
Variable Size             150995760 bytes
Database Buffers           16777216 bytes
Redo Buffers                2125824 bytes
SQL>

5、启动成功后，用rman恢复源机数据库的spfile

[oracle@hostb:/u01/oracle/product/db10gr2/dbs]$rman target /

Recovery Manager: Release 10.2.0.4.0 - Production on Wed Apr 27 16:01:32 2011

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

connected to target database: DEFAULT (not mounted)

RMAN> run {
2> ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' parms 'ENV=(NB_ORA_CLIENT=hosta)';
3> restore spfile from 'c-297293101-20110427-01';
4> release channel ch00;
5> }

using target database control file instead of recovery catalog
allocated channel: ch00
channel ch00: sid=47 devtype=SBT_TAPE
channel ch00: Veritas NetBackup for Oracle - Release 6.5 (2009050107)

Starting restore at 2011-04-27 16:03:59

channel ch00: autobackup found: c-297293101-20110427-01
channel ch00: SPFILE restore from autobackup complete
Finished restore at 2011-04-27 16:05:46

released channel: ch00

RMAN>

6、接下来将oracle shutdown，如果需要修改spfile可以先生成pfile文件进行修改然后在create spfile

第7和第8步用来恢复controlfile
7、使用新的spfile启动oracle到nomount状态

[oracle@hostb:/u01/oracle/admin/sxzhdb/bdump]$sqlplus /nolog

SQL*Plus: Release 10.2.0.4.0 - Production on Wed Apr 27 16:35:50 2011

Copyright (c) 1982, 2007, Oracle.  All Rights Reserved.

SQL> connect /as sysdba;
Connected to an idle instance.
SQL> startup nomount;
ORACLE instance started.

Total System Global Area 2.1475E+10 bytes
Fixed Size                  2107632 bytes
Variable Size            1996490512 bytes
Database Buffers         1.9462E+10 bytes
Redo Buffers               14667776 bytes
SQL>

8、启动没有问题的话就可以开始恢复controlfile了

[oracle@hfylhp01:/u01/oracle/admin/sxzhdb/bdump]$rman target /

Recovery Manager: Release 10.2.0.4.0 - Production on Wed Apr 27 16:39:09 2011

Copyright (c) 1982, 2007, Oracle.  All rights reserved.

connected to target database: sxzhdb (not mounted)

RMAN> run {
2> ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' parms 'ENV=(NB_ORA_CLIENT=hosta)';
restore controlfile from 'c-297293101-20110427-01';
3> 4> release channel ch00;
5> }

using target database control file instead of recovery catalog
allocated channel: ch00
channel ch00: sid=1091 devtype=SBT_TAPE
channel ch00: Veritas NetBackup for Oracle - Release 6.5 (2009050107)

Starting restore at 2011-04-27 16:39:32

channel ch00: restoring control file
channel ch00: restore complete, elapsed time: 00:01:57
output filename=/orasid/sys_u01/control01.ctl
output filename=/orasid/redo_u01/control02.ctl
output filename=/orasid/undo_u01/control03.ctl
Finished restore at 2011-04-27 16:41:29

released channel: ch00

RMAN>

9、确认controlfile恢复成功后将数据库启动到mount状态

RMAN> alter database mount;

database mounted

RMAN>

10、开始恢复数据库，这时最好写成脚本放在后台运行，并且将输出从定向到文件

[oracle@hostb:/usr/openv/scripts]$cat res.rcv
run {
ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' parms 'ENV=(NB_ORA_CLIENT=hosta)';
restore database;
recover database;
release channel ch00;
}
[oracle@hostb:/usr/openv/scripts]$cat res.sh
nohup rman nocatalog target / cmdfile=res.rcv msglog=res.log.`date +%y%m%d%H%M` &
[oracle@hostb:/usr/openv/scripts]$

直接执行res.sh，等到恢复完成后用命令打开数据库

alter database open resetlog;

完成