探索ORACLE之RMAN_07 数据库所有文件全部丢失恢复
作者：吴伟龙 Name：Prodence Woo
QQ：286507175 msn:hapy-wuweilong@hotmail.com
介于Rman恢复的自动性，我们刚才对数据库的全库，单表空间，数据文件控制文件和参数文件都做了备份，这些备份的内容是数据库运行不可缺少的必要元素，通常情况下我们在生产库中只需要备份这些内容即可。有了这些备份，即使数据库的的任何信息丢失，我们都可以得以恢复使得数据库的正常运行，当然如果我们将archivelog 也备份，当然就会更好了，一般情况下这个可以不用备份，备份archivelog占用空间非常大。如果有海量的储存，当然备份也无妨咯。
Egg:
删除数据库的所有文件，仅保留redo日志和archivelog文件
删除控制文件：
[oracle@wwldb WWL]$ pwd
/DBData/oradata/WWL
[oracle@wwldb WWL]$ ls
control02.ctl control03.ctl
[oracle@wwldb WWL]$ rm -rf *.ctl
[oracle@wwldbWWL]$
[oracle@wwldb WWL]$ pwd
/DBSoft/oradata/WWL
[oracle@wwldb WWL]$ ls
control01.ctl
[oracle@wwldb WWL]$ rm -rfcontrol01.ctl
[oracle@wwldbWWL]$
删除数据文件：
[oracle@wwldb WWL]$ ls
redo01.log redo02.log redo03.logsysaux01.dbf system01.dbf temp01.dbf undotbs01.dbf users01.dbf
[oracle@wwldb WWL]$ rm -rf *.dbf
[oracle@wwldb WWL]$ ls
redo01.log redo02.log redo03.log
[oracle@wwldbWWL]$
删除参数文件：
[oracle@wwldb dbs]$ cd$ORACLE_HOME/dbs
[oracle@wwldb dbs]$ rm -rfspfileWWL.ora
[oracle@wwldbdbs]$
执行大恢复:
1、现在数据库是关闭状态
[oracle@wwldb dbs]$ ps -ef|grep ora
root 2910 2888 0 10:02 ? 00:00:15hald-addon-storage: polling/dev/hdc
root 28954 3478 0 11:38 pts/100:00:00 su - oracle
oracle 28955 28954 0 11:38 pts/100:00:00 -bash
root 29361 3438 0 13:41 pts/200:00:00 su - oracle
oracle 29362 29361 0 13:41 pts/200:00:01 -bash
oracle 29988 29362 0 16:06 pts/200:00:00sqlplus as sysdba
oracle 30102 28955 0 17:03 pts/100:00:00 ps -ef
oracle 30103 28955 0 17:03 pts/100:00:00 grep ora
[oracle@wwldbdbs]$
2、要恢复首先要将数据库启动到mount状态才能恢复
[oracle@wwldb dbs]$ sqlplus / assysdba
SQL*Plus:Release 10.2.0.1.0 -Production on Wed May 23 17:03:42 2012
Copyright (c)1982, 2005, Oracle. Allrights reserved.
Connected to anidle instance.
SQL> startup
ORA-01078:failure in processingsystem parameters
LRM-00109: couldnot open parameterfile '/DBSoft/product/10.2.0/db_1/dbs/initWWL.ora'
SQL>
3、因为数据库没有参数文件，起不来，只能能启动到nomount状态
[oracle@wwldb dbs]$ rman target /
RecoveryManager: Release 10.2.0.1.0- Production on Wed May 23 17:06:32 2012
Copyright (c)1982, 2005, Oracle. Allrights reserved.
connected totarget database (notstarted)
RMAN> startup force nomount;
startup failed:ORA-01078: failure inprocessing system parameters
LRM-00109: couldnot open parameterfile '/DBSoft/product/10.2.0/db_1/dbs/initWWL.ora'
starting Oracleinstance withoutparameter file for retrival of spfile
Oracle instancestarted
Total System GlobalArea 159383552bytes
Fixed Size 1218244 bytes
VariableSize 58722620 bytes
DatabaseBuffers 92274688 bytes
RedoBuffers 7168000 bytes
RMAN>
4、开始恢复参数文件，只有恢复了参数文件和控制文件数据库才能到mount状态来恢复数据文件：
RMAN> restorespfile fromautobackup;
Starting restoreat 23-MAY-12
using targetdatabase control fileinstead of recovery catalog
allocatedchannel: ORA_DISK_1
channelORA_DISK_1: sid=36devtype=DISK
RMAN-00571:===========================================================
RMAN-00569:=============== ERRORMESSAGE STACK FOLLOWS ===============
RMAN-00571:===========================================================
RMAN-03002:failure of restorecommand at 05/23/2012 17:08:42
RMAN-06495: mustexplicitly specifyDBID with SET DBID command
RMAN>
注意：这里有问题了，spfile无法恢复，必须指定DBID。我记得在每次登录到rman的时候都会显示一个数据库的DBID，但是为什么这里要我们指定DBID呢？？？原因很简单，因为我们登录RMAN的时候，数据库已经宕机了，而且参数文件，控制文件，数据文件都没有了，它到哪里去找DBID；所以需要我们手动来指定，问题是现在既然登录的时候不显示，我们也不知道在哪里呀,如下:
[oracle@wwldb dbs]$ rman target /
Recovery Manager: Release 10.2.0.1.0- Production on WedMay 23 17:06:32 2012
Copyright (c) 1982, 2005, Oracle. All rightsreserved.
connected to targetdatabase (not started)
没关系的因为我们的控制文件做了自动备份，在自动备份的控制文件里面已经存在了DBID，我们现在去找找吧，默认就备份在如下位置：
[oracle@wwldb dbs]$ pwd
/DBSoft/product/10.2.0/db_1/dbs
[oracle@wwldb dbs]$ ls
c-5520179-20120518-01c-5520179-20120523-01 hc_WWL.dat lkWWL
其中5520179就是DBID了，我们只需要通过指定这个DBID，就可以将spfile恢复并将数据库启动到mount状态。
RMAN> set DBID=5520179
executingcommand: SET DBID
RMAN> restore spfile fromautobackup;
Starting restoreat 23-MAY-12
using targetdatabase control fileinstead of recovery catalog
allocatedchannel: ORA_DISK_1
channelORA_DISK_1: sid=39devtype=DISK
channelORA_DISK_1: looking forautobackup on day: 20120523
channelORA_DISK_1: autobackup found:c-5520179-20120523-04
channelORA_DISK_1: SPFILE restorefrom autobackup complete
Finished restoreat 23-MAY-12
RMAN>
5、恢复控制文件
RMAN> restore controlfile fromautobackup;
Starting restoreat 23-MAY-12
using channelORA_DISK_1
channelORA_DISK_1: looking forautobackup on day: 20120523
channelORA_DISK_1: autobackup found:c-5520179-20120523-04
channelORA_DISK_1: control filerestore from autobackup complete
outputfilename=/DBSoft/product/10.2.0/db_1/dbs/cntrlWWL.dbf
Finished restoreat 23-MAY-12
RMAN>
重启数据库到mount状态---很重要
SQL> startup mount;
ORACLE instancestarted.
Total SystemGlobal Area 285212672bytes
Fixed Size 1218968 bytes
VariableSize 79693416 bytes
DatabaseBuffers 197132288 bytes
RedoBuffers 7168000 bytes
Databasemounted.
SQL> exit
Disconnectedfrom Oracle Database 10gEnterprise Edition Release 10.2.0.1.0 - Production
With thePartitioning, OLAP and DataMining options
[oracle@wwldb dbs]$ rman target /
RecoveryManager: Release 10.2.0.1.0- Production on Wed May 23 19:04:15 2012
Copyright (c)1982, 2005, Oracle. Allrights reserved.
connected totarget database: WWL(DBID=5520179, not open)
注意：详细参考如下：
http://blog.csdn.net/wuweilong/article/details/7596056
http://space.itpub.net/?uid-20674423-action-viewspace-itemid-730717
6、恢复数据库其它的数据文件：
RMAN> restore database;
Starting restoreat 23-MAY-12
using targetdatabase control fileinstead of recovery catalog
allocatedchannel: ORA_DISK_1
channelORA_DISK_1: sid=156devtype=DISK
channelORA_DISK_1: starting datafilebackupset restore
channelORA_DISK_1: specifying datafile(s)to restore from backup set
restoringdatafile 00004 to/DBData/WWL/users01.dbf
channelORA_DISK_1: reading frombackup piece /DBBak/bak_WWL_05_23_0cnbn2a3_1_1
channelORA_DISK_1: restored backuppiece 1
piecehandle=/DBBak/bak_WWL_05_23_0cnbn2a3_1_1tag=TAG20120523T133147
channelORA_DISK_1: restore complete,elapsed time: 00:00:02
channelORA_DISK_1: starting datafilebackupset restore
channelORA_DISK_1: specifyingdatafile(s) to restore from backup set
restoringdatafile 00001 to/DBData/WWL/system01.dbf
restoringdatafile 00002 to/DBData/WWL/undotbs01.dbf
restoringdatafile 00003 to/DBData/WWL/sysaux01.dbf
channelORA_DISK_1: reading frombackup piece /DBBak/bak_WWL_05_23_0enbn2bk_1_1
channelORA_DISK_1: restored backuppiece 1
piecehandle=/DBBak/bak_WWL_05_23_0enbn2bk_1_1tag=TAG20120523T133235
channelORA_DISK_1: restore complete,elapsed time: 00:01:05
Finished restoreat 23-MAY-12
RMAN>
RMAN> recover database;
Starting recoverat 23-MAY-12
using channel ORA_DISK_1
starting mediarecovery
media recovery complete,elapsedtime: 00:00:05
Finished recoverat 23-MAY-12
RMAN> alter database open; ---------我们看到数据库已经打开了处于open状态
database opened
SQL> select instance_name,statusfrom v$instance;
INSTANCE_NAME STATUS
----------------------------
WWL OPEN
SQL>
