高可用 & 容灾16 分钟阅读
实战篇：Oracle 26ai DataGuard 搭建与 SwitchOver 切换
@TOC目录 这两天正好安装了 Oracle 26ai 的单机和 RAC 环境，想着正好可以搭建一套 DataGuard 环境玩玩，说干就干！ 数据库安装可参考以下两篇： - 《首发｜Oracle 26ai 本地安装包与
2026年1月30日阅读—点赞—收藏—
oracleracdataguard
内容阅读
在知识库中专注阅读，并随时返回相关工具与课程
前言
本章目标：掌握本章核心知识点 前置要求：完成前序章节学习 预计时长：60 分钟
这两天正好安装了 Oracle 26ai 的单机和 RAC 环境，想着正好可以搭建一套 DataGuard 环境玩玩，说干就干！
数据库安装可参考以下两篇：
本文不再赘述数据库安装相关内容。
ADG 搭建
依然采用使用 Duplicate 在线复制的方式搭建 ADG，无需 RMAN 备份，可以通过 RMAN Duplicate 直接在线从主库拷贝文件。
本文搭建步骤均来自 100天精通Oracle 实战系列 课程。
环境信息
hosts 配置
1# 所有节点配置 hosts 解析2cat << EOF >> /etc/hosts310.168.1.165 rac-scan410.168.1.25 orcldg5EOF主库配置
主库开启归档和强制日志
搭建 DataGuard 必须要开启归档模式和强制日志模式：
1-- 检查当前状态2SELECT log_mode, force_logging FROM v$database;34-- 开启归档模式（如未开启）5SHUTDOWN IMMEDIATE;6STARTUP MOUNT;7ALTER DATABASE ARCHIVELOG;8ALTER DATABASE OPEN;910-- 开启强制日志11ALTER DATABASE FORCE LOGGING;1213-- 确认配置生效14SELECT log_mode, force_logging FROM v$database;
配置 Standby Redo Logs
Standby Redo Logs 数量为 (Online Redo Logs + 1) × Threads：
1-- 查看在线日志配置2SELECT thread#, group#, bytes/1024/1024 size_mb, members 3FROM v$log 4ORDER BY thread#, group#;56-- 为每个线程添加Standby Redo Logs7-- Thread 1 (假设在线日志组数=7，大小=1GB)8ALTER DATABASE ADD STANDBY LOGFILE THREAD 19GROUP 20 ('+DATA', '+DATA') SIZE 1G,10GROUP 21 ('+DATA', '+DATA') SIZE 1G,11GROUP 22 ('+DATA', '+DATA') SIZE 1G,12GROUP 23 ('+DATA', '+DATA') SIZE 1G,13GROUP 24 ('+DATA', '+DATA') SIZE 1G,14GROUP 25 ('+DATA', '+DATA') SIZE 1G,15GROUP 26 ('+DATA', '+DATA') SIZE 1G,16GROUP 27 ('+DATA', '+DATA') SIZE 1G;1718-- Thread 219ALTER DATABASE ADD STANDBY LOGFILE THREAD 220GROUP 30 ('+DATA', '+DATA') SIZE 1G,21GROUP 31 ('+DATA', '+DATA') SIZE 1G,22GROUP 32 ('+DATA', '+DATA') SIZE 1G,23GROUP 33 ('+DATA', '+DATA') SIZE 1G,24GROUP 34 ('+DATA', '+DATA') SIZE 1G,25GROUP 35 ('+DATA', '+DATA') SIZE 1G,26GROUP 36 ('+DATA', '+DATA') SIZE 1G,27GROUP 37 ('+DATA', '+DATA') SIZE 1G;2829-- 验证Standby Redo Logs30SELECT thread#, group#, bytes/1024/1024 size_mb, status 31FROM v$standby_log 32ORDER BY thread#, group#;
生成并准备密码文件
rac 的密码文件通常存放于 ASM 盘中，查看对应的密码文件位置：
1srvctl config database -d orcl | grep PASSWORD
使用 asmcmd 进入到 ASM 磁盘组：
1# 从 ASM 复制到文件系统2asmcmd -p3ASMCMD [+] > pwcopy +DATA/RPTDB/PASSWORD/pwdrptdb.256.1180304871 /home/grid/
将密码文件拷贝到备库，并且重命名，命名规则为 orapw+sid：
1chown oracle:oinstall /home/grid/pwdorcl.256.12237618252mv /home/grid/pwdorcl.256.1223761825 /home/oracle/orapworcldg3scp /home/oracle/orapworcldg oracle@10.168.1.25:/u01/app/oracle/product/26.1.0/db/dbs/
生成参数文件
使用 RMAN 备份在备库恢复，需要在主库生成参数文件：
1-- 在主库生成参数文件2CREATE PFILE='/home/oracle/pfile_orcl.ora' FROM SPFILE;
备库配置
创建参数文件
将主库生成并修改后的参数文件在备库创建初始化参数文件（这里建议源端目标端均使用 OMF）：
1## 在备库创建初始化参数文件2cat<<-\EOF>$ORACLE_HOME/dbs/initorcldg.ora3*._b_tree_bitmap_plans=FALSE4*._cleanup_rollback_entries=20005*._optimizer_adaptive_cursor_sharing=FALSE6*._optimizer_extended_cursor_sharing='NONE'7*._optimizer_extended_cursor_sharing_rel='NONE'8*._optimizer_use_feedback=FALSE9*._undo_autotune=FALSE10*.compatible='23.6.0'11*.control_file_record_keep_time=3112*.control_files='/oradata/ORCL/CONTROLFILE/control01.ctl','/oradata/ORCL/CONTROLFILE/control02.ctl'13*.db_block_size=819214*.db_create_file_dest='/oradata'15*.db_files=500016*.db_name='orcl'17*.deferred_segment_creation=FALSE18*.diagnostic_dest='/u01/app/oracle'19*.dispatchers='(PROTOCOL=TCP) (SERVICE=orcldgXDB)'20*.enable_pluggable_database=true21*.event='28401 trace name context forever,level 1','10949 trace name context forever,level 1'22*.log_archive_format='%t_%s_%r.dbf'23*.nls_language='SIMPLIFIED CHINESE'24*.nls_territory='CHINA'25*.open_cursors=100026*.parallel_force_local=TRUE27*.parallel_max_servers=6428*.pga_aggregate_target=264346009629*.processes=200030*.remote_login_passwordfile='exclusive'31*.session_cached_cursors=30032*.sga_max_size=1057593753633*.sga_target=1057593753634*.undo_retention=1080035*.undo_tablespace='UNDOTBS1'36*.db_unique_name='orcldg'37*.fal_client='ORCL_STB'38*.fal_server='ORCL_PRI'39*.log_archive_config='dg_config=(orcl,orcldg)'40*.log_archive_dest_1='LOCATION=/oradata/archivelog VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=ORCLDG'41*.log_archive_dest_2='SERVICE=ORCL_PRI VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=ORCL'42*.standby_file_management='AUTO'43EOF创建目录结构
备库根据参数文件在两个节点均创建必要的目录：
1# 创建必要目录2mkdir -p /oradata/ORCL/{CONTROLFILE,DATAFILE,ONLINELOG}3mkdir -p /oradata/archivelog4chown -R oracle:oinstall /oradata5chmod -R 755 /oradata启动备库到 NOMOUNT
打开备库到 nomount 状态：
1-- 创建SPFILE2CREATE SPFILE FROM PFILE;34-- 启动到NOMOUNT状态5STARTUP NOMOUNT;
配置 TNS
备库启动监听：
1# 启动监听2lsnrctl start3## 等一段时间或者手动注册监听，再次查看监听状态4lsnrctl stat
配置 TNSNAMES.ORA：
1## 所有节点均配置 TNS2cat >> $ORACLE_HOME/network/admin/tnsnames.ora << 'EOF'3ORCL_PRI =4 (DESCRIPTION =5 (ADDRESS = (PROTOCOL = TCP)(HOST = 10.168.1.165)(PORT = 1521))6 (CONNECT_DATA =7 (SERVER = DEDICATED)8 (SERVICE_NAME = orcl)9 )10 )1112ORCL_STB =13 (DESCRIPTION =14 (ADDRESS = (PROTOCOL = TCP)(HOST = 10.168.1.25)(PORT = 1521))15 (CONNECT_DATA =16 (SERVER = DEDICATED)17 (SERVICE_NAME = orcldg)18 (UR=A)19 )20 )21EOF2223# 测试连接24tnsping ORCL_PRI25tnsping ORCL_STB2627# 验证数据库连接28sqlplus sys/oracle@ORCL_PRI as sysdba29sqlplus sys/oracle@ORCL_STB as sysdba
RMAN在线复制
rman 同时连接主备库：
1[oracle@orcldg:/home/oracle]$ rman target sys/oracle@ORCL_PRI AUXILIARY sys/oracle@ORCL_STB
执行 duplicate 命令：
1RMAN> run {2allocate channel prmy1 type disk;3allocate channel prmy2 type disk;4allocate channel prmy3 type disk;5allocate channel prmy4 type disk;6allocate channel prmy5 type disk;7allocate channel prmy6 type disk;8allocate auxiliary channel aux1 type disk;9allocate auxiliary channel aux2 type disk;10allocate auxiliary channel aux3 type disk;11allocate auxiliary channel aux4 type disk;12allocate auxiliary channel aux5 type disk;13allocate auxiliary channel aux6 type disk;14duplicate target database for standby from active database dorecover nofilenamecheck;15}
等待 RMAN duplicate 复制完成即可。
打开备库只读
1SQL> alter database open read only;2SQL> alter pluggable database all open;
备库开启日志应用
1SQL> alter database recover managed standby database using current logfile disconnect;23Database altered.
主库设置 ADG 参数
1alter system set log_archive_config='DG_CONFIG=(orcl,orcldg)';2alter system set log_archive_dest_1='LOCATION=+DATA VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=ORCL';3alter system set log_archive_dest_2='SERVICE=ORCL_STB VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=ORCLDG';4alter system set fal_client='ORCL_PRI';5alter system set fal_server='ORCL_STB';6alter system set log_archive_dest_state_2=defer;7alter system set log_archive_dest_state_2=enable;
检查同步情况
1-- 主库2SQL> set line2222 pages10003col status for a104col type for a105col error for a206col gap_status for a207col synchronization_status for a308col recovery_mode for a609select inst_id,status,DEST_ID,TYPE,ERROR,GAP_STATUS,SYNCHRONIZED,SYNCHRONIZATION_STATUS,RECOVERY_MODE from GV$ARCHIVE_DEST_STATUS where STatus <> 'INACTIVE' and type = 'PHYSICAL';
备库：
1SQL> set line222 pages10002select process,thread#,group#,sequence#,status from v$managed_standby;34SQL> col member for a1005select t2.thread#,t1.group#,t1.member,t2.STATUS,t2.ARCHIVED,t2.bytes/1024/1024 from v$logfile t1,v$standby_log t2 where t1.group#=t2.group# order by 1,2;
至此，ADG 搭建完成，备库可用于报表查询。
主备同步验证
在主库创建一个 PDB：
1-- 创建PDB2CREATE PLUGGABLE DATABASE testpdb ADMIN USER admin IDENTIFIED BY oracle;34-- 打开PDB5ALTER PLUGGABLE DATABASE testpdb OPEN;6ALTER PLUGGABLE DATABASE testpdb SAVE STATE;78-- 连接到新创建的 PDB9ALTER SESSION SET CONTAINER=testpdb;1011-- 创建测试用户12CREATE USER testuser IDENTIFIED BY oracle;13GRANT dba TO testuser;1415-- 创建测试表16CREATE TABLE testuser.employees (17 id NUMBER PRIMARY KEY,18 name VARCHAR2(50),19 department VARCHAR2(30),20 salary NUMBER,21 hire_date DATE22);2324-- 插入测试数据25INSERT INTO testuser.employees VALUES (1, 'John Doe', 'IT', 5000, SYSDATE-100);26INSERT INTO testuser.employees VALUES (2, 'Jane Smith', 'HR', 4500, SYSDATE-50);27INSERT INTO testuser.employees VALUES (3, 'Bob Johnson', 'Sales', 4000, SYSDATE-30);2829COMMIT;3031-- 验证数据32SELECT * FROM testuser.employees;33SELECT COUNT(*) FROM testuser.employees;
备库验证数据：
1-- 启动 PDB2SQL> alter pluggable database testpdb open;34-- 查看 PDB5SQL> show pdbs67-- 切换到 PDB8ALTER SESSION SET CONTAINER=testpdb;910-- 查询数据11SELECT * FROM testuser.employees;12SELECT COUNT(*) FROM testuser.employees;
主备 SwitchOver 切换
切换前检查
查看主备库同步进程状态：
1-- 主库2SQL> set line22223col status for a104col type for a105col error for a206col gap_status for a207col synchronization_status for a308col recovery_mode for a609select inst_id,status,DEST_ID,TYPE,database_mode,ERROR,GAP_STATUS,SYNCHRONIZED,SYNCHRONIZATION_STATUS,RECOVERY_MODE from GV$ARCHIVE_DEST_STATUS where STatus <> 'INACTIVE' and type = 'PHYSICAL';1011SQL> set line222212col name for a1013col database_role for a2014col switchover_status for a2015col PROTECTION_MODE for a2516select inst_id,name,database_role,protection_mode,switchover_status from gv$database;
备库：
1SQL> SELECT 2 (SELECT MAX(sequence#) FROM v$archived_log WHERE applied='YES') last_applied_seq,3 (SELECT MAX(sequence#) FROM v$archived_log) last_received_seq,4 (SELECT MAX(sequence#) FROM v$archived_log) - 5 (SELECT MAX(sequence#) FROM v$archived_log WHERE applied='YES') gap6FROM dual;78SQL> set line2229col database_role for a2010col switchover_status for a2011col PROTECTION_MODE for a2512select inst_id,name,database_role,protection_mode,switchover_status from gv$database;
主切换到备
主库切换为备库 (期间观察主库 alert 日志)：
1SQL> alter database commit to switchover to physical standby with session shutdown;
备切换到主
备库下检查是否可以切换为主库：
1SQL> set line2222col database_role for a203col switchover_status for a204col PROTECTION_MODE for a255select name,database_role,protection_mode,switchover_status from gv$database;
备库切换主库：
1SQL> alter database commit to switchover to primary with session shutdown;23Database altered.
启动主备库
打开新的主库：
1SQL> alter database open;23Database altered.45SQL> select name,open_mode,database_role from v$database;67NAME OPEN_MODE DATABASE_ROLE8--------- -------------------- ----------------9ORADB READ WRITE PRIMARY
打开新的备库：
1[oracle@rac01:/home/oracle]$ srvctl start db -d orcl23SQL> alter database recover managed standby database using current logfile disconnect from session;4SQL> select name,open_mode,database_role from gv$database;
检查同步情况
检查主备同步情况：
1-- 主库2SQL> set line22223col status for a104col type for a105col error for a206col gap_status for a207col synchronization_status for a308col recovery_mode for a609select inst_id,status,DEST_ID,TYPE,database_mode,ERROR,GAP_STATUS,SYNCHRONIZED,SYNCHRONIZATION_STATUS,RECOVERY_MODE from GV$ARCHIVE_DEST_STATUS where STatus <> 'INACTIVE' and type = 'PHYSICAL';
切换后主备同步正常，操作流程也较为简单。
总结
这篇文章是使用经典的架构去搭建和切换 ADG，所以还是比较顺利，但是我看 26ai 支持 PDB 级别的 ADG，后面有时间再学习一下。