1、副总裁需要裸恢复的严峻现实
集团总部的信息部负责人给我打电话说为了找一年前的记录，所以需要对一年前2015年5月1日的数据进行恢复。而2016年初因为进行迁移，所以有些文件可能丢失，手上只有rman全备文件，希望在一天之内找回，集团一个副总裁在等着这个数据有急用。
我在电话里面说马上去做，接完电话，想到只有rman备份文件，而且是备份的数据文件，没有控制文件没有参数文件的备份，所以普通的
（1）先恢复控制文件restore controlfile from ‘…bak’;
（2）然后catalog start with ‘/data/2015-05-01/’注册备份文件，
（3）最后restore database;recover database;恢复数据库。
三板斧的常规途径是彻底的行不通了，咋办？咋办？咋办呢？……
2、先进行数据文件的剥离
突然想起以前记得看过关于restoreDatafileTo数据抽取的操作思路，大概是如果没有控制文件后，可以从rman的数据文件备份和归档日志备份里面抽取数据文件，然后重新建立控制文件，再用resetlogs方式打开数据库：
因为抽取命令里面需要填写一个个数据文件，这里有个前提是自己要熟悉自己的oracle实例的文件目录，知道备份的时候oracle实例有多少个数据文件（包括文件存放目录），这样就可以快速的整理出来抽取命令的sql。如果这些都忘记了，还可以去备份日志文件里面去查看check下，一般备份日志文件里面都会有的。
根据以前记录整理下抽取命令：
DECLARE devtype varchar2(256); done boolean; BEGIN devtype:=sys.dbms_backup_restore.deviceAllocate (type=>'',ident=>'t1'); sys.dbms_backup_restore.restoreSetDatafile; sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>01,toname=>'/home/oradata/powerdes/system01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>02,toname=>'/home/oradata/powerdes/sysaux01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>03,toname=>'/home/oradata/powerdes/undotbs01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>04,toname=>'/home/oradata/powerdes/users01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>05,toname=>'/home/oradata/powerdes/powerdesk01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>06,toname=>'/home/oradata/powerdes/plas01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>07,toname=>'/home/oradata/powerdes/pl01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>08,toname=>'/home/oradata/powerdes/help01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>09,toname=>'/home/oradata/powerdes/adobelc01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>10,toname=>'/home/oradata/powerdes/sms01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>11,toname=>'/home/oradata/powerdes/plcrm01.dbf'); sys.dbms_backup_restore.restoreBackupPiece(done=>done,handle=>'/data/2015-05-01/full_POWERDES_20150501_3566.bak',params=>null); sys.dbms_backup_restore.deviceDeallocate; END; / |
执行过程如下：
[oracle@pldb236 oradata]$ rlwrap sqlplus / as sysdba sql*Plus: Release 11.2.0.1.0 Production on Mon Oct 17 21:19:32 2016 Copyright (c) 1982,2009,Oracle. All rights reserved. Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production With the Partitioning,OLAP,Data Mining and Real Application Testing options DECLARE owerdes/system01.dbf'); sys.dbms_backup_restore.restoreDatafileTo(dfnumber=>02,toname=>'/home/oradata/powerdes/undotbs01.dbf'); devtype varchar2(256); done boolean; BEGIN devtype:=sys.dbms_backup_restore.deviceAllocate (type=>'',params=>null); sys.dbms_backup_restore.deviceDeallocate; END; 21 / PL/sql procedure successfully completed. sql> |
后台alert日志会显示正在不停的剥离出文件到指定目录里面去：
| CKPT started with pid=13,OS id=23993 Mon Oct 17 21:16:59 2016 SMON started with pid=14,OS id=23995 Mon Oct 17 21:16:59 2016 RECO started with pid=15,OS id=23997 Mon Oct 17 21:16:59 2016 MMON started with pid=16,OS id=23999 starting up 1 dispatcher(s) for network address '(ADDRESS=(PARTIAL=YES)(PROTOCOL=TCP))'... Mon Oct 17 21:16:59 2016 MMNL started with pid=17,OS id=24001 starting up 1 shared server(s) ... ORACLE_BASE from environment = /oracle/app/oracle Mon Oct 17 21:20:14 2016 Full restore complete of datafile 7 to datafile copy /home/oradata/powerdes/pl01.dbf. Elapsed time: 0:00:27 checkpoint is 11106141982 last deallocation scn is 11082135537 Full restore complete of datafile 8 to datafile copy /home/oradata/powerdes/help01.dbf. Elapsed time: 0:00:03 checkpoint is 11106141982 last deallocation scn is 9881798870 Full restore complete of datafile 9 to datafile copy /home/oradata/powerdes/adobelc01.dbf. Elapsed time: 0:00:01 checkpoint is 11106141982 Mon Oct 17 21:20:32 2016 Full restore complete of datafile 10 to datafile copy /home/oradata/powerdes/sms01.dbf. Elapsed time: 0:00:10 checkpoint is 11106141982 Mon Oct 17 21:21:25 2016 Full restore complete of datafile 3 to datafile copy /home/oradata/powerdes/undotbs01.dbf. Elapsed time: 0:01:23 checkpoint is 11106141982 last deallocation scn is 11106022955 Undo Optimization current scn is 11106076830 Mon Oct 17 21:22:28 2016 Full restore complete of datafile 4 to datafile copy /home/oradata/powerdes/users01.dbf. Elapsed time: 0:02:53 checkpoint is 11106141982 last deallocation scn is 11082633897 Mon Oct 17 21:23:47 2016 Full restore complete of datafile 11 to datafile copy /home/oradata/powerdes/plcrm01.dbf. Elapsed time: 0:04:11 checkpoint is 11106141982 last deallocation scn is 11100156728 Mon Oct 17 21:25:30 2016 Full restore complete of datafile 1 to datafile copy /home/oradata/powerdes/system01.dbf. Elapsed time: 0:05:34 checkpoint is 11106141982 last deallocation scn is 11039454999 Undo Optimization current scn is 11106076830 Mon Oct 17 21:28:10 2016 Full restore complete of datafile 2 to datafile copy /home/oradata/powerdes/sysaux01.dbf. Elapsed time: 0:08:27 checkpoint is 11106141982 last deallocation scn is 11101587434 Mon Oct 17 21:29:11 2016 Full restore complete of datafile 6 to datafile copy /home/oradata/powerdes/plas01.dbf. Elapsed time: 0:09:33 checkpoint is 11106141982 last deallocation scn is 11082142314 |
3、建立控制文件
数据文件抽取成功后，需要单独自己创建控制文件，如果不知道如何创建controlfile的命令，可以在线上生成trace文件一般默认的控制文件是二进制的，打开来是乱码的 ，备份一个trace出来 可以打开看到语句了,$ORACLE_BASE/admin/$ORACLE_SID/udump目录下，生成的新的 trace 文件里，trace文件有生成控制文件的脚本，使用如下命令alter database backup controlfile to trace as'/oracle/app/oracle/admin/powerdes/pfile/control.sql';可以得到创建控制文件的sql命令。
这里有个前提是自己要熟悉自己的oracle实例的文件目录，知道备份的时候oracle实例有多少个数据文件，有多少个redo log文件，这样就可以快速的整理出来创建控制文件的sql。如果这些都忘记了，还可以去备份日志文件里面去查看check下，一般备份日志文件里面都会有的。
（1）命令如下 CREATE CONTROLFILE REUSE SET DATABASE "POWERDES" RESETLOGS ARCHIVELOG MAXLOGFILES 16 MAXLOGMEMBERS 3 MAXDATAFILES 100 MAXINSTANCES 8 MAXLOGHISTORY 2920 DATAFILE '/home/oradata/powerdes/system01.dbf', '/home/oradata/powerdes/sysaux01.dbf', '/home/oradata/powerdes/undotbs01.dbf', '/home/oradata/powerdes/users01.dbf', '/home/oradata/powerdes/powerdesk01.dbf', '/home/oradata/powerdes/plas01.dbf', '/home/oradata/powerdes/pl01.dbf', '/home/oradata/powerdes/help01.dbf', '/home/oradata/powerdes/adobelc01.dbf', '/home/oradata/powerdes/sms01.dbf', '/home/oradata/powerdes/plcrm01.dbf' LOGFILE GROUP 1 '/home/oradata/powerdes/redo01.log' SIZE 50M BLOCKSIZE 512, GROUP 2 '/home/oradata/powerdes/redo02.log' SIZE 50M BLOCKSIZE 512, GROUP 3 '/home/oradata/powerdes/redo03.log' SIZE 50M BLOCKSIZE 512 CHARACTER SET ZHS16GBK; （2）执行过程如下： sql> CREATE CONTROLFILE REUSE SET DATABASE "POWERDES" RESETLOGS ARCHIVELOG MAXLOGFILES 16 MAXLOGMEMBERS 3 MAXDATAFILES 100 MAXINSTANCES 8 MAXLOGHISTORY 2920 DATAFILE '/home/oradata/powerdes/system01.dbf', GROUP 3 '/home/oradata/powerdes/redo03.log' SIZE 50M BLOCKSIZE 512 23 CHARACTER SET ZHS16GBK; Control file created. sql> |
4、进行数据恢复
开始执行数据恢复，还是依然在sql窗口里面进行操作的，操作如下：
#(1) 执行recover恢复 sql> recover database using backup controlfile until cancel ; ORA-00279: change 11106141982 generated at 05/01/2015 03:00:08 needed for thread 1 ORA-00289: suggestion : /oracle/app/oracle/flash_recovery_area/POWERDES/archivelog/2016_10_17/o1_mf_1_32 117_%u_.arc ORA-00280: change 11106141982 for thread 1 is in sequence #32117 Specify log: {<RET>=suggested | filename | AUTO | CANCEL} # 这里一般选择输入cancel即可 cancel ORA-10879: error signaled in parallel recovery slave ORA-01547: warning: RECOVER succeeded but OPEN RESETLOGS would get error below ORA-01194: file 2 needs more recovery to be consistent ORA-01110: data file 2: '/home/oradata/powerdes/sysaux01.dbf' sql> alter database open resetlogs; alter database open resetlogs * ERROR at line 1: ORA-01194: file 2 needs more recovery to be consistent ORA-01110: data file 2: '/home/oradata/powerdes/sysaux01.dbf' # 这时，我们无法将数据库打开，一直报ORA-01194错误，说明数据库的SCN号和数据文件的SCN号不一致了。这是因为控制文件我们不是从备份集里面恢复回来的，而是在抽取数据文件后手动建立的控制文件，因此要比数据文件的SCN号要大（甚至特殊情况当前的数据库的会是0）。通过对v$database和v$datafile的checkpoint_change#列的查询，可以确定出本次操作中当前数据库的checkpoint_chenage#为0，两者完全不一致导致通过resetlogs打开数据库异常。
Bty：如果这里当前数据库v$database的值不为0，但是仍然比数据文件v$datafile里面的值大，那么则会不停报ORA-01152错误。 sql> select checkpoint_change# from v$database; CHECKPOINT_CHANGE# ------------------ 0 sql> select file#,checkpoint_change# from v$datafile; FILE# CHECKPOINT_CHANGE# ---------- ------------------ 1 1.1106E+10 2 1.1106E+10 3 1.1106E+10 4 1.1106E+10 5 1.1106E+10 6 1.1106E+10 7 1.1106E+10 8 1.1106E+10 9 1.1106E+10 10 1.1106E+10 11 1.1106E+10 11 rows selected. sql> |
文章源地址：http://www.jb51.cc/article/p-sbirgpvl-bau.html，未经过作者mchdba（黄杉）允许，谢绝转载。
怎么办呢？这个时候，就需要我们使用_allow_resetlogs_corruption的隐含参数来处理了。
整个调整的目标是强制启动数据库，设置此参数之后，在数据库Open过程中，Oracle会跳过某些一致性检查，从而使数据库可能跳过不一致状态，Open打开:
# 启动隐含参数 sql> alter system set "_allow_resetlogs_corruption"=true scope=spfile; System altered. sql> # 然后重启数据库，使参数生效，在此基础上再次恢复数据库 sql> shutdown immediate; ORA-01109: database not open Database dismounted. ORACLE instance shut down. sql> startup mount; ORACLE instance started. Total System Global Area 6680915968 bytes Fixed Size 2213936 bytes Variable Size 4898949072 bytes Database Buffers 1744830464 bytes Redo Buffers 34922496 bytes Database mounted. sql> recover database using backup controlfile until cancel; ORA-00279: change 11106141982 generated at 05/01/2015 03:00:08 needed for thread 1 ORA-00289: suggestion : /oracle/app/oracle/flash_recovery_area/POWERDES/archivelog/2016_10_17/o1_mf_1_32 117_%u_.arc ORA-00280: change 11106141982 for thread 1 is in sequence #32117 Specify log: {<RET>=suggested | filename | AUTO | CANCEL} cancel ORA-10879: error signaled in parallel recovery slave ORA-01547: warning: RECOVER succeeded but OPEN RESETLOGS would get error below ORA-01194: file 2 needs more recovery to be consistent ORA-01110: data file 2: '/home/oradata/powerdes/sysaux01.dbf' # 然后使用resetlogs打开数据库，成功了。 sql> alter database open resetlogs; Database altered. sql> # 然后检查scn，都是统一的了。 select checkpoint_change# from v$database; select file#,checkpoint_change# from v$datafile; |
然后使用业务表数据来判断是否已经恢复成功到这一天，查看后确认成功：
sql> select t2.* from(select t.uiid,t.updated_date from plas.plas_acct t where t.updated_date is not null order by t.updated_date desc ) t2 where rownum <10; UIID UPDATED_DATE -------------------------------------------------- ------------ wangyu1 01-MAY-15 gaihy 01-MAY-15 xuhl 01-MAY-15 dingchuan1 01-MAY-15 zhangcong 01-MAY-15 chenwh2 01-MAY-15 yuli2 01-MAY-15 zhangxya 01-MAY-15 qiuwj 01-MAY-15 9 rows selected. sql> |
至此，没有控制文件下通过rman恢复一年的数据做成了，然后通过expdp导出需要的数据，之后对数据库进行恢复或者重建等等。
原文链接：https://www.f2er.com/oracle/212212.html