Oracle 10g Dataguard实施

一、实施最大性能的物理Standby

1、先建立主库orcl

 

2、主库需要配置db_unique_name参数，如果从来没有配置过，需要停库进行修改。

 

确认SPFILE中

 

DB_UNIQUE_NAME=orcl1

DB_NAME=orcl

LOG_ARCHIVE_CONFIG='dg_confg=(orcl1, orcl2)'     #即log_archive_config='dg_config=(主库唯一名,从库唯一名)'

LOG_ARCHIVE_MAX_PROCESSES=4;      #适当增加归档进程数量

LOG_ARCHIVE_DEST_1='location=/arc1 VALID_FOR=(ONLINE_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=orcl1'

LOG_ARCHIVE_DEST_2='service=orcl2 VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=orcl2'

3、在从库上建立参数文件

*.background_dump_dest='/u01/oracle/admin/orcl2/bdump'

*.core_dump_dest='/u01/oracle/admin/orcl2/cdump'

*.user_dump_dest='/u01/oracle/admin/orcl2/udump'

*.compatible='10.2.0.1.0'

*.db_name=orcl

*.db_unique_name='orcl2'

*.control_files='/u01/app/oracle/oradata/orcl1/control01.ctl'

*.REMOTE_LOGIN_PASSWORDFILE=EXCLUSIVE

*.db_block_size=8192

*.sga_target=160m

*.pga_aggregate_target=20m

*.undo_management='AUTO'

*.undo_tablespace='UNDOTBS1'

*.standby_file_management='AUTO'

*.log_archive_config='dg_config=(orcl1,orcl2)'

*.log_archive_dest_10='location=/arch2 VALID_FOR=(ONLINE_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=orcl2'

*.standby_archive_dest='/arc1'

#在oracle 11g r2里standby_archive_dest已经废弃

*.db_file_name_convert=('/u01/app/oracle/oradata/orcl/','/u01/app/oracle/oradata/orcl2/')

*.log_file_name_convert=('/u01/app/oracle/oradata/orcl/','/u01/app/oracle/oradata/orcl2/')

 

4、从主库拷贝口令文件

scp host1:/u01/app/oracle/product/10.2.0/db_1/dbs/orapworcl host2:/u01/app/oracle/product/10.2.0/db_1/dbs/orapworcl2

 

5、配置net

netmgr

静态注册orcl1，orcl2

 

6、备份主库

rman target /

CONFIGURE CHANNEL DEVICE TYPE DISK FORMAT '/home/oracle/orcl1/%d_%I_%s_%p.bkp';

BACKUP AS COMPRESSED BACKUPSET DATABASE INCLUDE CURRENT CONTROLFILE FOR STANDBY PLUS ARCHIVELOG;

 

7、拷贝备份片到从库

scp host1:/home/oracle/orcl1/* host2:/home/oracle/orcl2/

 

8、在主库使用RMAN对备库进行还原

rman target sys/oracle@orcl1 auxiliary sys/oracle@orcl2

DUPLICATE TARGET DATABASE FOR STANDBY;

 

执行结束后，从库会被启动到MOUNT模式，数据同步是ARCHIVELOG文件级别的。

9、其它步骤

注意：如果没有在从库配置

*.db_file_name_convert=('/u01/app/oracle/oradata/orcl/','/u01/app/oracle/oradata/orcl2/')

*.log_file_name_convert=('/u01/app/oracle/oradata/orcl/','/u01/app/oracle/oradata/orcl2/')

则有可能报RMAN-05001: auxiliary file name/u01/app/oracle…… conflicts with a file used by thetarget database错误

此时需要DUPLICATE TARGET DATABASE FOR STANDBY NOFILENAMECHECK;

确认环节：（在主库中执行）

col dest_name for a30

col error for a20

select dest_name,status,error,target,process from v$archive_dest;

查看一下结果集的各目标状态是否正常。

从库进入管理恢复模式：

ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT FROM SESSION;

ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;

在这之后可以将从库open

ALTER DATABASE OPEN;

10、将DataGuard从“最大性能模式”升级成为“最高可用模式（最高可用物理standby）”

（1）主库调整LOG粒度

ALTER SYSTEM SET log_archive_dest_1='service=orcl2 lgwr sync affirm VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=orcl2'

（2）添加standby logfile用来接收从主库传过来的LOG

ALTER DATABASE ADD STANDBY LOGFILE 'XXX' SIZE 50M;

ALTER DATABASE ADD STANDBY LOGFILE 'XXX' SIZE 50M;

ALTER DATABASE ADD STANDBY LOGFILE 'XXX' SIZE 50M;

ALTER DATABASE ADD STANDBY LOGFILE 'XXX' SIZE 50M;

建议比主库的LOGFILE多一组

（3）设置从库的归档路径

ALTER SYSTEM SET log_archive_dest_3='location=/arc3 valid_for=(STANDBY_LOGFILES,STANDBY_ROLE) db_unique_name=orcl2'

（4）主库设置为最高可用模式

alter database set standby database to maximize availability;

（5）确认数据库保护模式

select protection_mode,database_role,protection_level from v$database;

（6）从库进入管理恢复模式

开启物理standby的日志实时应用(从库):

alter database recover managed standby database using current logfile disconnect from session;

停止日志应用:

alter database recover managed standby database cancel;

打开数据库

alter database open;

这样就可以用从库充当主数据库使用了。