Oracle数据库工程师手记:从RAC环境备份后向新环境(文件系统)恢复的试验

2022-08-19 21:08:59 浏览数 (1)

编者按:

本文作者系大连健哥, POSTGRESQL、ORACLE 数据库资深从业人员、IT 技术的深度爱好者。相信科学改变人类、技术创造未来。个人主页:https://www.cnblogs.com/gaojian/,经其本人授权发布。

【免责声明】本号文章仅代表个人观点,与任何公司无关。

假定备份源是RAC ASM环境,执行恢复操作的对象服务器是文件系统环境。假定备份源端具有一个称为 /home/grid/mnt的挂载点作为NFS客户端,在还原/恢复目标端,作为 NFS服务器的对应目录是 /home/oracle/formnt。

我的备份脚本如下:

Level 0 增量备份:

代码语言:javascript复制
run{
 
backup incremental level 0 format ' DATA' database;
sql 'alter system archive log current';
backup  as backupset format ' DATA' archivelog from time 'sysdate-1';
backup format ' DATA' current controlfile;
 
}

Level 1 增量备份:

代码语言:javascript复制
run{
 
  backup incremental level 1 format ' DATA' database;
  sql 'alter system archive log current';
  backup as backupset format ' DATA' archivelog from time 'sysdate-1';
  backup format ' DATA' current controlfile;
  }
 

仅使用0级和1级备份可能无法还原到指定的时间点。因此,上面的脚本中,切换 archive log 并对其进行备份。在执行备份之前,记录一下数据文件和 Online REDO 文件路径。

代码语言:javascript复制
SQL> select name from v$datafile;
 
 NAME
 --------------------------------------------------------------------------------
  DATA/ORA12102/DATAFILE/system.261.943312827
  DATA/ORA12102/DATAFILE/sysaux.273.838525691
  DATA/ORA12102/DATAFILE/undotbs1.276.998325422
  DATA/ORA12102/DATAFILE/users.262.987321245     
 
 SQL> select member from gv$logfile;
 
 MEMBER
 --------------------------------------------------------------------------------
  DATA/ORA12102/ONLINELOG/group_3.264.954612469
  DATA/ORA12102/ONLINELOG/group_3.263.954612505
  DATA/ORA12102/ONLINELOG/group_2.266.954612377
  DATA/ORA12102/ONLINELOG/group_2.265.954612423
  DATA/ORA12102/ONLINELOG/group_1.268.954612297
  DATA/ORA12102/ONLINELOG/group_1.267.954612333
 
 SQL>

Source 端的备份经过:

Level 0 增量备份

代码语言:javascript复制
RMAN> run{
  backup incremental level 0 format ' DATA' database;
  sql 'alter system archive log current';
  backup  as backupset format ' DATA' archivelog from time 'sysdate-1';
  backup format ' DATA' current controlfile;
}

模拟数据库的运行:变更一些数据

代码语言:javascript复制
SQL> create table tab001 as select * from dba_objects;
SQL> alter system archive log current;

Level 1 增量备份

代码语言:javascript复制
RMAN> run{
 backup incremental level 1 format ' DATA' database;
 sql 'alter system archive log current';
 backup as backupset format ' DATA' archivelog from time 'sysdate-1';
 backup format ' DATA' current controlfile;
}

检查备份的结果:

代码语言:javascript复制
$ export NLS_DATE_FORMAT='YYYYMMDD_HH24:MI:SS'
$ rman target /
 
RMAN> list backup;
 
BS Key  Type LV Size   Device Type Elapsed Time Completion Time
------- ---- -- ---------- ----------- ------------ -----------------
5   Incr 0  1.20G  DISK00:04:50 20210407_12:07:23
 BP KEY: 5   STATUS: AVAILABLE  Compressed : NO  TAG: TAG20210407T120233
 PIECE NAME:  DATA/ORA12102/BACKUPSET/2017_09_14/sssdr0_tag20210407t120233_0.260.954676953
…..
File LV Type Ckp SCNCkp Time Name
---- -- ---- ---------- ----------------- ----
10  Incr 184119020210407_12:02:33  DATA/ORA12102/DATAFILE/system.261.943312827
30  Incr 184119020210407_12:02:33  DATA/ORA12102/DATAFILE/sysaux.273.838525691
40  Incr 184119020210407_12:02:33  DATA/ORA12102/DATAFILE/undotbs1.276.998325422
60  Incr 184119020210407_12:02:33  DATA/ORA12102/DATAFILE/users.262.987321245     
……
 
BS Key  Type LV Size   Device Type Elapsed Time Completion Time
------- ---- -- ---------- ----------- ------------ ---------------
6  Full10.19M DISK00:00:01 20210407_12:07:23
BP Key: 34   Status: AVAILABLE  Compressed: NO  Tag: TAG: TAG20210407T120233
Piece Name: /refresh/home/app/12.1.0.2/oracle/product/12.2.0.1/dbhome3/dbs/ncsnn1_tag20210407t125241_0.280.954680049
  SPFILE Included: Modification time: 20210407_12:07:23
  SPFILE db_unique_name: ORA12102
  Control File Included: Ckp SCN: 1950495  Ckp time: 20210407_12:07:23
<<<<  此处 "SPFILE Include" 是随着 level 0 增分备份执行自动生成的包含 controlfile 和 spfile 信息的备份
 
BS Key  Type LV Size   Device Type Elapsed Time Completion Time
------- ---------- ----------- ------------ -----------------
11  31.47M DISK00:00:13 20210407_12:54:35
 BP KEY:11   STATUS: AVAILABLE  Compressed: NO  TAG: TAG20210407T125422
 PIECE NAME:  DATA/ORA12102/BACKUPSET/2017_09_14/annnf0_tag20210407t125422_0.283.954680063
 
BS Key  Type LV Size   Device Type Elapsed Time Completion Time
Thrd Seq Low SCNLow TimeNext SCN   Next Time
---- ------- ---------- ----------------- ---------- ---------
114  183759620210407_11:22:53 184050920210407_11:48:07
115  184050920210407_11:48:07 184133520210407_12:07:41
116  184133520210407_12:07:41 184134320210407_12:07:43
117  184134320210407_12:07:43 184266320210407_12:51:19
118  184266320210407_12:51:19 184276220210407_12:54:19
119  184276220210407_12:54:19 184277020210407_12:54:20    <===请记住最新 archive log 的 Next SCN,因为我们将在还原时使用这个值。
 
BS Key  Type LV Size   Device Type Elapsed Time Completion Time
------- ---- -- ---------- ----------- ------------ -----------------
12  Full9.61M  DISK00:00:07 20210407_12:54:44
 BP KEY: 12   STATUS: AVAILABLE  COMPRESSION: NO  TAG: TAG20210407T125437
 PIECE NAME :  DATA/ORA12102/BACKUPSET/2017_09_14/ncnnf0_tag20210407t125437_0.284.954680081
Controlfile included : Ckp SCN: 1842784  Ckp time: 20210407_12:54:37
 
RMAN>

使用asmcmd通过NFS将source 端ASM上的备份复制到dest 端。假设此NFS挂载点是 grid 用户主目录下的mnt目录。执行以下操作:

代码语言:javascript复制
[root@ol66 ~]# su - grid
[grid@ol66 ~]$ export ORACLE_SID= ASM
 
ASMCMD>cp  DATA/ORA12102/BACKUPSET/2017_09_14/sssdr0_tag20210407t120233_0.260.954676953  /home/grid/mnt
......

将上面 <list backup> 中列出的 备份文件,都拷贝到 dest 端(过程略)。

Dest 端执行恢复操作:

代码语言:javascript复制
# su - oracle
$ . oraenv
ORACLE_SID = [ora12102] ?   <<<<<<<<<<<<<<<<<  指定 Source 端的 ORACLE_SID
ORACLE_HOME = [/home/oracle] ? /u01/app/oracle/product/12.1.0/dbhome_1   <<<<<<<<<<<<<<<<<<< 设定 ORACLE_HOME(根据当初安装 Oracle binary 时的设定来指定)
The Oracle base remains unchanged with value /u01/app/oracle
$

利用前面的SPFILE 的备份(ncsnn1_tag20210407t125241_0.280.954680049),来在 dest 端恢复 SPFILE。

代码语言:javascript复制
$ rman target /
 
RMAN> startup nomount force;
 
Failed to start: ORA-01078: failure in processing system parameters
LRM-00109: ??????????????????????????????'/u01/app/oracle/product/12.1.0/dbhome_1/dbs/initora12102.ora'?????????????????????????????????
...

RMAN>

不必在意上面的错误,继续执行:

代码语言:javascript复制
RMAN> restore spfile from '/home/oracle/formnt/ncsnn1_tag20210407t125241_0.280.954680049';

也作成 PFILE(是为了读取spfile 中的内容,并做出相应更改)

代码语言:javascript复制
RMAN> restore spfile to pfile '/home/oracle/initora12102.ora' from '/home/oracle/formnt/ncsnn1_tag20210407t125241_0.280.954680049';
 
RMAN> exit

由于source 端是RAC环境,dest 端是 single 环境,因此将 cluster_database设置为false或删除PFILE中的该行。然后检查有关PFILE中每个目录的信息。

代码语言:javascript复制
$ cat  initora12102.ora


ora12102.__data_transfer_cache_size=0
ora12102.__db_cache_size=1040187392
ora12102.__java_pool_size=16777216
ora12102.__large_pool_size=33554432
ora12102.__oracle_base='/u01/app/oracle'#ORACLE_BASE set from environment
ora12102.__pga_aggregate_target=503316480
ora12102.__sga_target=1493172224
ora12102.__shared_io_pool_size=67108864
ora12102.__shared_pool_size=318767104
ora12102.__streams_pool_size=0
*.audit_file_dest='/u01/app/oracle/admin/ora12102/adump'
*.audit_trail='db'
*.compatible='12.1.0.2.0'
*.control_files=' DATA/ORA12102/CONTROLFILE/current.269.954612287',' DATA/ORA12102/CONTROLFILE/current.272.954612291'
*.db_block_size=8192
*.db_create_file_dest=' DATA'
*.db_domain=''
*.db_name='ora12102'
*.db_recovery_file_dest=' DATA'
*.db_recovery_file_dest_size=4560m
*.diagnostic_dest='/u01/app/oracle'
*.dispatchers='(PROTOCOL=TCP) (SERVICE=ora12102XDB)'
*.open_cursors=300
*.pga_aggregate_target=470m
*.processes=300
*.remote_login_passwordfile='EXCLUSIVE'
*.sga_target=1410m
*.undo_tablespace='UNDOTBS1'
$

参照上面的内容,在 dest 端,建立好目录。或者修改 pfile 里关于给中目录的信息,建立好对应的目录。

代码语言:javascript复制
mkdir -p /u01/app/oracle/admin/ora12102/adump

因为 dest 端是文件系统,所以没有磁盘组,把涉及到磁盘组的内容进行修改,并建立相应目录(如:控制文件所在目录)

代码语言:javascript复制
mkdir -p /u01/app/oracle/fast_recovery_area/CONTROLFILE
 
mkdir -p /u01/app/oracle/oradata/CONTROLFILE

为在文件系统存储数据文件,建立相应的目录。

代码语言:javascript复制
mkdir -p /u01/app/oracle/oradata/datafile

为了在文件系统上存储 online redo,也建立好相应的目录。

代码语言:javascript复制
mkdir -p /u01/app/oracle/oradata/logfile

由于不能使用 ASM磁盘组,所以调整哪些用到磁盘组的路径,变成使用文件系统路径(如 control_files的值)。

代码语言:javascript复制
ora12102.__data_transfer_cache_size=0
ora12102.__db_cache_size=1040187392
ora12102.__java_pool_size=16777216
ora12102.__large_pool_size=33554432
ora12102.__oracle_base='/u01/app/oracle'#ORACLE_BASE set from environment
ora12102.__pga_aggregate_target=503316480
ora12102.__sga_target=1493172224
ora12102.__shared_io_pool_size=67108864
ora12102.__shared_pool_size=318767104
ora12102.__streams_pool_size=0
*.audit_file_dest='/u01/app/oracle/admin/ora12102/adump'
*.audit_trail='db'
*.compatible='12.1.0.2.0'
*.control_files='/u01/app/oracle/fast_recovery_area/CONTROLFILE/control01.ctl','/u01/app/oracle/oradata/CONTROLFILE/control02.ctl'
*.db_block_size=8192
*.db_create_file_dest='/u01/app/oracle/oradata'
*.db_domain=''
*.db_name='ora12102'
*.db_recovery_file_dest='/u01/app/oracle/fast_recovery_area'
*.db_recovery_file_dest_size=4560m
*.diagnostic_dest='/u01/app/oracle'
*.dispatchers='(PROTOCOL=TCP) (SERVICE=ora12102XDB)'
*.open_cursors=300
*.pga_aggregate_target=470m
*.processes=300
*.remote_login_passwordfile='EXCLUSIVE'
*.sga_target=1410m
*.undo_tablespace='UNDOTBS1'

现在 PFILE 修改好了,可以再次启动了:

代码语言:javascript复制
$ sqlplus / as sysdba
 
SQL> shu immediate
SQL> startup nomount pfile='/home/oracle/initora12102.ora'
 
SQL> exit

用 RMAM 恢复 控制文件:

代码语言:javascript复制
$ rman target /
 
RMAN> restore controlfile from '/home/oracle/formnt/ncnnf0_tag20210407t125437_0.284.954680081';
RMAN> exit

Source 端,使用的是 ASM 磁盘组,dest 端使用文件系统,所以,设置 db_file_name_convert 和 log_file_name_convert。

(用到我上面建立好的 目录 /u01/app/oracle/oradata/datafile/ 和 /u01/app/oracle/oradata/logfile/ )

代码语言:javascript复制
$ sqlplus / as sysdba
SQL>create spfile from pfile='/home/oracle/initora12102.ora';
 
SQL>shutdown immediate
 
SQL>startup mount
 
SQL> alter system set db_file_name_convert=' DATA/ORA12102/DATAFILE/','/u01/app/oracle/oradata/datafile/' scope=spfile;
SQL> alter system set log_file_name_convert=' DATA/ORA12102/ONLINELOG/','/u01/app/oracle/oradata/logfile/' scope=spfile;
 
SQL> shutdown immediate
 
SQL> startup mount
SQL> exit

由于在 source 端时,备份建立到了 磁盘组上,控制文件中也是如此记忆的。所以需要在 恢复控制文件之后,重新登记这些 备份的位置。

代码语言:javascript复制
$ rman target /
 
 RMAN>
 
 RMAN> crosscheck backup;
 RMAN> delete expired backup;
 RMAN> catalog start with '/home/oracle/formnt/';  <<<<<<<<  上述拷贝好的备份的位置。

执行数据库的 restore / recover :

代码语言:javascript复制
RMAN> run{
   set until scn 1842770;   <<<<<<<  这个  SCN 的值,就是我们在 上面 list backup 时看到的最新的 archive log 的 next scn 值   
   restore database;
   recover database;
}

打开数据库。

代码语言:javascript复制
RMAN> alter database open resetlogs;
 
RMAN>exit

确认一下新数据库总的 vdatafile 和 vlogfile 中 数据文件和 online redo 内容。

代码语言:javascript复制
SQL> select name from v$datafile;
SQL> select member from v$logfile;
SQL> exit   

至此,向新环境的恢复操作结束。

0 人点赞