Oracle 10G windows 平台 DataGuard 实例

server A: 10.85.10.44SID=orcl ­

target server B: 10.85.10.43 ­

oracle version:10.2.0.1 ­

OS platform : windows XP ­

­

­

Primary设置： ­

­

1. 设置主数据库为force logging 模式 ­

SQL>sqlplus "/as sysdba" ­

SQL>alter database force logging; ­

2. 设置主数据库为归档模式 ­

SQL> archive log list ­

SQL> shutdown immediate ­

SQL> startup mount ­

SQL> alter database archivelog; ­

SQL> archive log list ­

3. 添加"备用联机日志文件" ­

先查看日志文件位置： ­

SQL>select * from v$logfile; ­

在添加： ­

SQL> alter database add standby logfile group 4 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo04.log') size 50m; ­

SQL> alter database add standby logfile group 5 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo05.log') size 50m; ­

SQL> alter database add standby logfile group 6 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo06.log') size 50m; ­

SQL> alter database add standby logfile group 7 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo07.log') size 50m; ­

­

4. 创建主库的初始化参数给备库用 ­

SQL>Create pfile from spfile; ­

产生的文件名为initorcl.ora 存放目录默认放在$ORACLE_HOME/database下 ­

­

5. 在主库创建备库的控制文件 和密码文件 ­

SQL>Alter database create standby controlfile as 'D:/oracle/product/10.2.0/control01.ctl'; ­

D:>orapwd file=D:/PWDorcl.ora password=admin entries=5; ­

缺省情况下，win下口令文件的格式是pwdsid.ora，unix下的格式是orapwSID(大小写敏感）

­

6.在主库创建监听和配置tnsnams.ora ­

­

listener.ora配置如下: ­

# listener.ora Network Configuration File: D:/oracle/product/10.2.0/db_1/network/admin/listener.ora­

# Generated by Oracle configuration tools.­

SID_LIST_LISTENER =­

(SID_LIST =­

(SID_DESC =­

(SID_NAME = PLSExtProc)­

(ORACLE_HOME = D:/oracle/product/10.2.0/db_1)­

(PROGRAM = extproc)­

)­

(SID_DESC =­

(GLOBAL_DBNAME = orcl)­

(ORACLE_HOME = D:/oracle/product/10.2.0/db_1)­

(SID_NAME = orcl)­

)­

)­

LISTENER =­

(DESCRIPTION_LIST =­

(DESCRIPTION =­

(ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC1))­

(ADDRESS = (PROTOCOL = TCP)(HOST = 10.85.3.44)(PORT = 1521))­

)­

)­

­

tnsnames.ora配置如下: ­

# tnsnames.ora Network Configuration File: D:/oracle/product/10.2.0/db_1/network/admin/tnsnames.ora­

# Generated by Oracle configuration tools.­

primary =­

(DESCRIPTION =­

(ADDRESS = (PROTOCOL = TCP)(HOST = 10.85.3.44)(PORT = 1521))­

(CONNECT_DATA =­

(SERVER = DEDICATED)­

(SERVICE_NAME = orcl)­

)­

)­

standby =­

(DESCRIPTION =­

(ADDRESS = (PROTOCOL = TCP)(HOST = 10.85.3.43)(PORT = 1521))­

(CONNECT_DATA =­

(SERVER = DEDICATED)­

(SERVICE_NAME = orcl)­

)­

)­

EXTPROC_CONNECTION_DATA =­

(DESCRIPTION =­

(ADDRESS_LIST =­

(ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC1))­

)­

(CONNECT_DATA =­

(SID = PLSExtProc)­

(PRESENTATION = RO)­

)­

)­

­

­

7. 在initorcl.ora添加以下内容： ­

*.log_archive_format='%T%S%r.ARC' ­

*.DB_UNIQUE_NAME='primary' ­

*.log_archive_config='DG_CONFIG=(primary,standby)' ­

*.log_archive_dest_1='location=D:/oracle/product/10.2.0/oradata/arch VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=primary' ­

*.log_archive_dest_2='SERVICE=standby arch ASYNC VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=standby' ­

*.STANDBY_FILE_MANAGEMENT=AUTO ­

*.LOG_ARCHIVE_DEST_STATE_1=ENABLE ­

*.LOG_ARCHIVE_DEST_STATE_2=ENABLE ­

*.FAL_SERVER='standby' ­

*.FAL_CLIENT='primary' ­

­

关闭数据库，在用initorcl.ora重启，并创建spfile文件 ­

SQL>startup pfile='$ORACLE_HOME/database/initorcl.ora' ­

SQL>create spfile from pfile='$ORACLE_HOME/database/initorcl.ora' ­

­

8.­

A. 拷贝数据文件，参数文件，控制文件，密码文件到备库上 ­

initorcl.ora参数文件,PWDorcl.ora密码文件考到$ORACLE_HOME/database下的， ­

controlbak.ctl 考到$ORACLE_base/oradata/orcl/下，并分别重命名为control01.ctl,control01.ctl,control01.ctl ­

数据文件考到$ORACLE_base/oradata/orcl/下 ­

­

B. 用Rman拷贝,不用停机­

$ rman target /­

RMAN> backup full format 'D:/FULL_%d_%T_%s.bak' database include current controlfile for standby;­

RMAN> sql 'alter system archive log current';­

RMAN> Backup ArchiveLog all format='D:/arch_%d_%T_%s.bak'; ­

备份完后将备份文件拷到standby上同样的目录，强调：同样的目录，在standby进行rman 恢复即可­

­

9. 启动主数据库 ­

SQL>startup ­

­

--------------------------------------------------------------------- ­

Standy操作： ­

­

1. 用oradim工具创建备库orcl实例 ­

oradim.exe -new -sid orcl -startmode m ­

oradim.exe -edit -sid orcl -startmode a ­

2. 创建备库存放数据文件和后台跟踪目录 ­

$ORACLE_BASE/ORADATA/ORCL ­

$ORACLE_BASE/admin/orcl ­

$ORACLE_BASE/admin/orcl/adump ­

$ORACLE_BASE/admin/orcl/bdump ­

$ORACLE_BASE/admin/orcl/cdump ­

$ORACLE_BASE/admin/orcl/dpdump ­

$ORACLE_BASE/admin/orcl/pfile ­

$ORACLE_BASE/admin/orcl/udump ­

$ORACLE_BASE/admin/orcl/ ­

3. 添加"备用联机日志文件" ­

SQL>startup mount ­

先查看日志文件位置： ­

SQL>select * from v$logfile; ­

在添加： ­

SQL> alter database add standby logfile group 4 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo04.log') size 50m; ­

SQL> alter database add standby logfile group 5 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo05.log') size 50m; ­

SQL> alter database add standby logfile group 6 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo06.log') size 50m; ­

SQL> alter database add standby logfile group 7 ('D:/ORACLE/PRODUCT/10.2.0/ORADATA/ORCL/redo07.log') size 50m; ­

­

4. 在备库创建监听和配置tnsnams.ora ­

­

listener.ora配置如下: ­

# listener.ora Network Configuration File: D:/oracle/product/10.2.0/db_1/network/admin/listener.ora­

# Generated by Oracle configuration tools.­

SID_LIST_LISTENER =­

(SID_LIST =­

(SID_DESC =­

(SID_NAME = PLSExtProc)­

(ORACLE_HOME = D:/oracle/product/10.2.0/db_1)­

(PROGRAM = extproc)­

)­

(SID_DESC =­

(GLOBAL_DBNAME = orcl)­

(ORACLE_HOME = D:/oracle/product/10.2.0/db_1)­

(SID_NAME = orcl)­

)­

)­

LISTENER =­

(DESCRIPTION_LIST =­

(DESCRIPTION =­

(ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC1))­

(ADDRESS = (PROTOCOL = TCP)(HOST = 10.85.3.44)(PORT = 1521))­

)­

)­

­

tnsnames.ora配置如下: ­

# tnsnames.ora Network Configuration File: D:/oracle/product/10.2.0/db_1/network/admin/tnsnames.ora­

# Generated by Oracle configuration tools.­

primary =­

(DESCRIPTION =­

(ADDRESS = (PROTOCOL = TCP)(HOST = 10.85.3.44)(PORT = 1521))­

(CONNECT_DATA =­

(SERVER = DEDICATED)­

(SERVICE_NAME = orcl)­

)­

)­

standby =­

(DESCRIPTION =­

(ADDRESS = (PROTOCOL = TCP)(HOST = 10.85.3.43)(PORT = 1521))­

(CONNECT_DATA =­

(SERVER = DEDICATED)­

(SERVICE_NAME = orcl)­

)­

)­

EXTPROC_CONNECTION_DATA =­

(DESCRIPTION =­

(ADDRESS_LIST =­

(ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC1))­

)­

(CONNECT_DATA =­

(SID = PLSExtProc)­

(PRESENTATION = RO)­

)­

)­

­

­

5. 测试主备之间网络连通 ­

primary: ­

C:>lsnrctl start ­

C:>tnsping standby ­

standby ­

C:>lsnrctl start ­

C:>tnsping primary ­

­

6. 配置备库初始化参数 ­

编辑$ORACLE_HOME/database目录下的initorcl.ora添加以下内容 ­

*.log_archive_format='%T%S%r.ARC' ­

*.DB_UNIQUE_NAME='standby' ­

*.log_archive_config='DG_CONFIG=(primary,standby)' ­

*.log_archive_dest_1='location=D:/oracle/product/10.2.0/oradata/arch VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=standby' ­

*.log_archive_dest_2='SERVICE=primary arch ASYNC VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=primary' ­

*.STANDBY_FILE_MANAGEMENT=AUTO ­

*.LOG_ARCHIVE_DEST_STATE_1=ENABLE ­

*.LOG_ARCHIVE_DEST_STATE_2=ENABLE ­

*.FAL_SERVER='primary' ­

*.FAL_CLIENT='standby' ­

­

7. 启动备用数据库 ­

C:>set ORACLE_SID=orcl ­

SQL>sqlplus "/as sysdba" ­

SQL>startup nomount pfile='D:/oracle/product/10.2.0/db_1/database/initorcl.ora'; ­

SQL>create spfile from pfile='D:/oracle/product/10.2.0/db_1/database/initorcl.ora'; ­

­

若采用Rman备份的, 则在此standby 端进行 Rman还原数据库：­

$rman target sys/admin@primary auxiliary /­

RMAN> duplicate target database for standby dorecover nofilenamecheck;

­

SQL>alter database mount standby database ; ­

SQL>alter database recover managed standby database disconnect from session; ­

­

--------------------------------------------------------------------------- ­

测试 ­

注意DataGuard启动顺序：

启动顺序：先standby,后primary;
关闭顺序：先primary后standby;

在备库将实例启动到mount状态：

SQL>startupnomount;

SQL>alterdatabasemountstandbydatabase;

SQL>ALTERDATABASERECOVERMANAGEDSTANDBYDATABASECANCEL;

SQL>alterdatabaserecovermanagedstandbydatabasedisconnectfromsession;

在备库启动监听：

$lsnrctlstart

在主库启动实例：

SQL>startup;

在主库启动监听：

$lsnrctlstart

在主库验证归档目录是否有效：

SQL>SELECTSTATUS,DESTINATION,ERRORFROMV$ARCHIVE_DEST;

如果有错误，要排查原因。

SQL>altersystemswitchlogfile;

SQL>selectmax(sequence#)fromv$archived_log;

MAX(SEQUENCE#)

--------------

70

主备查询结果一致，DataGuard搭建结束。

<!--EndFragment-->

­

1. 测试主库产生的归档日志是否能正常传送到归档日志 ­

主库进行日志切换: ­

SQL>Alter system switch logfile; ­

然后分别查看主库和备库的D:/arch目录下是否产生了同样的归档日志 ­

文件。 ­

select max(sequence#) from v$archived_log; ­

select max(sequence#) from v$log_history; ­

select group#,sequence#,archived,status from v$log; ­

select name,sequence#,applied from v$archived_log; ­

select sequence#,applied from v$archived_log; ­

­

若不同步， ­

1. 看log日志， archive是否有丢失 ­

2. 可以在备库坐如下操作： ­

alter database recover managed standby database cancel; ­

alter database recover managed standby database disconnect from session; ­

­

----------------------------------------------------------------------- ­

主备库切换 ­

­

1. switchover ­

­

一般SWITCHOVER切换都是计划中的切换，特点是在切换后，不会丢失任何的数据，而且这个过程是可逆的，整个DATA GUARD环境不会被破坏，原来DATA GUARD环境中的所有物理和逻辑STANDBY都可以继续工作。 在进行DATA GUARD的物理STANDBY切换前需要注意： ­

确认主库和从库间网络连接通畅； ­

确认没有活动的会话连接在数据库中； ­

PRIMARY数据库处于打开的状态，STANDBY数据库处于MOUNT状态； ­

确保STANDBY数据库处于ARCHIVELOG模式； ­

如果设置了REDO应用的延迟，那么将这个设置去掉； ­

确保配置了主库和从库的初始化参数，使得切换完成后，DATA GUARD机制可以顺利的运行。 ­

­

主库： ­

1. 查看switchover 状态 ­

SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE; ­

to standby ­

附： A:switchover_status出现session active/not allowed ­

当出现session active的时候表示还有活动的session,则运行 ­

Alter database commit to switchover to physical standby with session shutdown; ­

当出现not allowed时，在官方文档说转换会不成功，但是我测试的时候成功了，如果大家在测试不成功的时候再和我说，让我看看在什么情况下会不成功。 ­

­

B.ora- 01153: an incompatible media recovery is active ­

运行下面代码 ­

Alter database recover managed standby database finish; ­

或者Alter database recover managed standby database finish force; ­

Alter database recover managed standby database disconnect from session; ­

2 切换成备库 ­

SQL>Alter database commit to switchover to physical standby with session shutdown; ­

SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PHYSICAL STANDBY; ­

Database altered. ­

­

3 启动到mount和应用日志状态 ­

SQL> SHUTDOWN IMMEDIATE ­

SQL> startup nomount; ­

SQL> alter database mount standby database; ­

SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT FROM SESSION; ­

­

4. 查看数据库模式 ­

SQL>select dest_name,status,database_mode,recovery_mode,protection_mode from v$archive_dest_status; ­

SQL>select status,database_mode from v$archive_dest_status; ­

­

备库： ­

­

1.查看switchover状态 ­

SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE; ­

TO PRIMARY ­

附：若不是用此语句切换：ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY with session shutdown ­

2. 切换成主库 ­

SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY; ­

Database altered. ­

SQL> shutdown immediate; ­

SQL> startup; ­

SQL> alter system switch logfile; ­

3. 查看数据库模式 ­

SQL>select dest_name,status,database_mode,recovery_mode,protection_mode from v$archive_dest_status; ­

SQL>select status,database_mode from v$archive_dest_status; ­

­

注意地方: ­

如果做了switchover,主库参数设置成以下方式，会触发ora-16009错误 ­

Alert system set log_archive_dest_2=’service=primary ­

DB_UNIQUE_NAME=orcl’ scope=spfile; ­

然后再alert_orcl.log 日志中会出现以下错误内容 ­

Thu Nov 27 10:19:12 2008 ­

Redo Shipping Client Connect ­

-- Connected User is Valid ­

RFS[2]: Assigned to RFS process 1292 ­

RFS[2]: Database mount ID mismatch [0x4781d95f:0x47823be1] ­

RFS[2]: Client instance is standby database i ­

RFS[2]: Not using real app ­

Thu Nov 27 10:19Errors in file ­

d:/oracle/product/10.2.0/admin/orcl/udump/orc ­

ORA-16009: 远程归档日 ­

从metalink上查到: ­

* fact: Oracle Server - Enterprise Edition 9 ­

* symptom: Errors appears in alert.log on primary database ­

* symptom: RFS: client instance is standby database instead ­

* symptom: RFS: Not using real application clusters ­

* symptom: Errors appear in alert.log on standby database ­

* symptom: ­

database ­

standby database ­

primary database ­

* symptom: Standby redo log files are defined on the standby database ­

* cause: The standby redo log files are synchronously filled with redo ­

from the primary database. When a logswitch occur on the primary database, ­

those files are archived on the standby database before being applyed on ­

it. The archiving process on the standby database should only archive to ­

the local disks on tprimarfix: ­

Disable the remote archiving on the standby databasExample: alter system set log_archive_dest_2 = '' ­

是因为没有把standby 上的log_archive_dest_2 清空导致的。 ­

另外也有可 ­

bug 4676659 ­

Standby may not be recognised (ORA-16009) ­

When the log transport is LGWR ASYNC and logical standby has ­

LOG_ARCHIVE_DEST setting VALID_FOR=(ONLINE_LOGFILE, PRIMARY_ROLE) ­

ORA-16009 is reportedregular interWorkaround: ­

There is no workaround to prevent ORA-16009 from appearing in alert logs. ­

­

2. Failovers： ­

FAILOVER切换一般是PRIMARY数据库发生故障后的切换，这种情况是STANDBY数据库发挥其作用的情况。这种切换发生后，可能会造成数据的丢失。而且这个过程不是可逆的，DATA GUARD环境会被破坏。 ­

由于PRIMARY数据库已经无法启动，所以FAILOVER切换所需的条件并不多，只要检查STANDBY是否运行在最大保护模式下，如果是的话，需要将其置为最大性能模式，否则切换到PRIMARY角色也无法启动。 ­

­

­

1.查看是否有日志GAP,没有应用的日志： ­

SQL> SELECT UNIQUE THREAD#, MAX(SEQUENCE#) OVER(PARTITION BY THREAD#) LAST FROM V$ARCHIVED_LOG; ­

　　SQL> SELECT THREAD#, LOW_SEQUENCE#, HIGH_SEQUENCE# FROM V$ARCHIVE_GAP; ­

　　如果有，则拷贝过来并且注册 ­

SQL> ALTER DATABASE REGISTER PHYSICAL LOGFILE '路径'; ­

重复查看直到没有应用的日志： ­

2. 然后停止应用归档： ­

　　SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL; ­

　　Database altered. ­

3. 下面将STANDBY数据库切换为PRIMARY数据库： ­

　　SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE FINISH; ­

或 SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE FINISH FORCE; ­

　　Database altered. ­

　　SQL> SELECT DATABASE_ROLE FROM V$DATABASE; ­

　　DATABASE_ROLE ­

　　---------------- ­

　　PHYSICAL STANDBY ­

　　SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY; ­

　　Database altered. ­

SQL> ALTER DATABASE OPEN; 或者 shutdown immediate+startup ­

　　Database altered. ­

­

　　检查数据库是否已经切换成功： ­

　　SQL> SELECT DATABASE_ROLE FROM V$DATABASE; ­

　　DATABASE_ROLE ­

　　---------------- ­

　　PRIMARY ­

　　至此，FAILOVER切换完成。这个时候应该马上对新的PRIMARY数据库进行备份。­

数据库模式查看：

1.首先查看当前的保护模式---primary数据库操作

SQL>selectprotection_mode,protection_levelfromv$database;

PROTECTION_MODEPROTECTION_LEVEL

----------------------------------------

MAXIMUMPERFORMANCEMAXIMUMPERFORMANCE

2.设置新的数据保护模式并重启数据库--primary数据库操作

当保护模式更改顺序：

maximizeprotection--->maximizeavailability---->maximizeperformance

当在把dataguard的保护级别按这上面的顺序减低的时候，不需要primary库在mount状态，否则primary必须在mount状态。

如：
SQL>alterdatabasesetstandbydatabasetomaximizeavailability;

alterdatabasesetstandbydatabasetomaximizeavailability

*

ERRORatline1:

ORA-01126:databasemustbemountedinthisinstanceandnotopeninany

instance

SQL>shutdownimmediate

Databaseclosed.

Databasedismounted.

ORACLEinstanceshutdown.

SQL>startupmount;

ORACLEinstancestarted.

TotalSystemGlobalArea167772160bytes

FixedSize1218316bytes

VariableSize79694068bytes

DatabaseBuffers83886080bytes

RedoBuffers2973696bytes

Databasemounted.

SQL>alterdatabasesetstandbydatabasetomaximizeavailability;

Databasealtered.

SQL>alterdatabaseopen;

alterdatabaseopen

*

ERRORatline1:

ORA-03113:end-of-fileoncommunicationchannel

报错了，这是因为最大可用性需要先修改日志传送方式为lgwr同步方式，否则，数据库是无法open.

Maximumprotection/AVAILABILITY模式必须满足以下条件

RedoArchivalProcess:LGWR

NetworkTranmissionmode:SYNC

DiskWriteOption:AFFIRM

StandbyRedoLogs:Yes

standbydatabasetype:PhysicalOnly

SQL>altersystemsetlog_archive_dest_2='service=orcl_stlgwrsyncAFFIRM';

Systemaltered.

SQL>shutdownimmediate

Databaseclosed.

Databasedismounted.

ORACLEinstanceshutdown.

SQL>startupmount;

ORACLEinstancestarted.

TotalSystemGlobalArea167772160bytes

FixedSize1218316bytes

VariableSize79694068bytes

DatabaseBuffers83886080bytes

RedoBuffers2973696bytes

Databasemounted.

SQL>alterdatabasesetstandbydatabasetomaximizeavailability;

Databasealtered.

SQL>alterdatabaseopen;

Databasealtered.

SQL>selectprotection_mode,protection_levelfromv$database;

PROTECTION_MODEPROTECTION_LEVEL

----------------------------------------

MAXIMUMAVAILABILITYMAXIMUMAVAILABILITY

alterdatabasesetstandbydatabasetomaximizeperformance;

提示：maximize后可跟{PROTECTION|AVAILABILITY|PERFORMANCE}，分别对应最大保护，最高可用性及最高性能。

在最大保护模式下，直接关闭备库是不行的，如果在备库上关闭数据库，会有如下提示：
SQL>shutdownimmediate
ORA-01154:databasebusy.Open,close,mount,anddismountnotallowednow
SQL>
在最大保护模式下，备库是不允许关闭的，此时首先关闭主库，然后备库就可以顺利关闭了。

注意：主库的保护模式修改之后，备库的模式也会改变，和主库保持一致。

<!--EndFragment-->