给客户搭建RAC-DG时碰到的一些问题总结

昨天去某客户那里搭建RAC-DG的物理备库，这里把大致流程再列举一下,为不涉及泄露隐私，主要参数已替换名称，具体路径也不一致。由于客户那边的环境不允许本机用网络连接内网，因此无法截下故障时各种报警日志和trace文件的截图，只能凭回忆大致写出。 1.主库

昨天去某客户那里搭建RAC-DG的物理备库，这里把大致流程再列举一下,为不涉及泄露隐私，主要参数已替换名称，具体路径也不一致。由于客户那边的环境不允许本机用网络连接内网，因此无法截下故障时各种报警日志和trace文件的截图，只能凭回忆大致写出。

1.主库在线修改spfile参数

alter database force logging;

alter system set log_archive_cOnfig=&＃39;DG_COnFIG=(dg,dgdg)&＃39;;
alter system set log_archive_dest_1=&＃39;LOCATION=USE_DB_RECOVERY_FILE_DEST VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=dg&＃39;;
alter system set log_archive_dest_2=&＃39;SERVICE=dgdg LGWR ASYNC VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=dgdg&＃39;;
alter system set log_archive_dest_state_1=&＃39;ENABLE&＃39;;
alter system set log_archive_dest_state_2=&＃39;ENABLE&＃39;;
alter system set standby_file_management=&＃39;AUTO&＃39;;

alter system set fal_server=&＃39;DGDG&＃39;;

需重启参数：

alter system set db_file_name_cOnvert=&＃39;C:\app\administrator\oradata\dg&＃39;,&＃39;+data/dg/datafile&＃39; scope=spfile;
alter system set log_file_name_cOnvert=&＃39;C:\app\administrator\oradata\dg&＃39;,&＃39;+data/dg/onlinelog&＃39; scope=spfile;

2.创建pfile

SQL> create pfile from spfile;

3.修改备库用的pfile

*.__db_cache_size=0

*.__java_pool_size=0
*.__large_pool_size=0
*.__oracle_base=&＃39;C:\app\administrator&＃39;#ORACLE_BASE set from environment
*.__pga_aggregate_target=0
*.__sga_target=0
*.__shared_io_pool_size=0
*.__shared_pool_size=0
*.__streams_pool_size=0
*.audit_file_dest=&＃39;C:\app\administrator\admin\dg\adump&＃39;
*.audit_trail=&＃39;db&＃39;
*.cluster_database=false --说明是单实例数据库，否则启动会报错
*.compatible=&＃39;11.2.0.0.0&＃39;
*.control_files=&＃39;C:\app\administrator\oradata\dg\control01.ctl&＃39;,&＃39;C:\app\administrator\oradata\dg\control02.ctl&＃39;
*.db_block_size=8192
*.db_create_file_dest=&＃39;C:\app\administrator\oradata&＃39;
*.db_domain=&＃39;&＃39;
*.db_name=&＃39;dg&＃39;
*.db_recovery_file_dest=&＃39;C:\app\administrator\flash_recovery_area&＃39;
*.db_recovery_file_dest_size=3908042752
*.diagnostic_dest=&＃39;C:\app\administrator&＃39;
*.dispadghers=&＃39;(PROTOCOL=TCP) (SERVICE=DGXDB)&＃39;
*.fal_server=&＃39;dg&＃39;
*.instance_number=1
*.log_archive_cOnfig=&＃39;DG_COnFIG=(dg,dgdg)&＃39;
*.log_archive_dest_1=&＃39;LOCATION=c:\archivelog VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=dgdg&＃39;
*.log_archive_dest_2=&＃39;SERVICE=dg LGWR ASYNC VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=dg&＃39;
*.log_archive_dest_state_1=&＃39;ENABLE&＃39;
*.log_archive_dest_state_2=&＃39;ENABLE&＃39;
*.log_archive_format=&＃39;ARC%S_%R.%T&＃39;
*.memory_target=8577351680
*.nls_language=&＃39;SIMPLIFIED CHINESE&＃39;
*.nls_territory=&＃39;CHINA&＃39;
*.open_cursors=300
*.processes=150
*.remote_login_passwordfile=&＃39;exclusive&＃39;
*.standby_file_management=&＃39;AUTO&＃39;
*.undo_tablespace=&＃39;UNDOTBS1&＃39;
*.db_file_name_cOnvert=&＃39;+data/dg/datafile&＃39;,&＃39;C:\app\administrator\oradata\dg&＃39;
*.log_file_name_cOnvert=&＃39;+data/dg/onlinelog&＃39;,&＃39;C:\app\administrator\oradata\dg&＃39;
*.db_unique_name=dgdg --复制过来的pfile如果不设置这个参数，默认的值是dg

4.修改tnsnamesl.ora

# tnsnames.ora Network Configuration File: C:\app\administrator\product\11.2.0\dbhome_1\network\admin\tnsnames.ora

# Generated by Oracle configuration tools.

dg =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = dg-cluster-scan)(PORT = 1521))
(CONNECT_DATA =
(SERVER = DEDICATED)
(SERVICE_NAME = dg)
)
)


DGDG =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = dgP)(HOST = dgdg)(PORT = 1521))
)
(CONNECT_DATA =
(SERVICE_NAME = dg)
)
)

5.修改listener.ora(只给备库用，主库可以不配置静态监听)

# listener.ora Network Configuration File: C:\app\administrator\product\11.2.0\dbhome_1\NETWORK\ADMIN\listener.ora
# Generated by Oracle configuration tools.

SID_LIST_LISTENER =
(SID_LIST =
(SID_DESC =
(GLOBAL_DBNAME = dg)
(ORACLE_HOME = C:\app\administrator\product\11.2.0\dbhome_1)
(SID_NAME = dg)
)
)

LISTENER =
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP)(HOST = dgdg)(PORT = 1521))
)

ADR_BASE_LISTENER = C:\app\administrator\product\11.2.0\dbhome_1\log

6.备库创建相关目录

c:\archivelog --指定一个本地归档路径，备库接收到的归档日志和自己生成的归档日志都放在这里
c:\app\administrator\admin\dg\adump
c:\app\administrator\admin\dg\dpdump
c:\app\administrator\admin\dg\hdump
c:\app\administrator\admin\dg\pfile
c:\app\administrator\flash_recovery_area
c:\app\administrator\oradata\dg

7.主库做rman全备

RMAN> backup as compressed backupset full database format &＃39;c:\bak\full_%d_%I_%T_%U&＃39;

8.主库创建备库控制文件

SQL> alter database create standby controlfile as &＃39;c:\control01.ctl&＃39;;
SQL> alter database create standby controlfile as &＃39;c:\control02.ctl&＃39;;

9.复制备份文件、密码文件、pfile文件、tnsnames.ora、listener.ora到备库相应位置

10.备库创建实例

oradim -new -sid dg -startmode manual -spfile;

11.启动监听

lsntrctl start

12.启动实例到mount

set oracle_sid=dg
sqlplus / as sysdba
SQL> startup mount

13.恢复数据库

RMAN> catalog start with &＃39;d:\bak&＃39;; --不指定会提示无法恢复数据库
RMAN> restore database;

14.备库添加standby redo logfile

SQL> alter database add standby logfile &＃39;C:\app\administrator\oradata\dg\std_05.log&＃39; size 50m;
SQL> alter database add standby logfile &＃39;C:\app\administrator\oradata\dg\std_06.log&＃39; size 50m;
SQL> alter database add standby logfile &＃39;C:\app\administrator\oradata\dg\std_07.log&＃39; size 50m;
SQL> alter database add standby logfile &＃39;C:\app\administrator\oradata\dg\std_08.log&＃39; size 50m;
SQL> alter database add standby logfile &＃39;C:\app\administrator\oradata\dg\std_09.log&＃39; size 50m;

15.启用redo apply

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

16.给备库创建spfile(可选)

SQL> create spfile from pfile;

下面记录几个在整个配置过程中遇到的问题：

1.用opatch apply命令无法打patch

出现原因：11.2.0.3默认装完后的opatch版本是11.2.0.1.7,我要打的patch 27需要在这个版本之上才可以

解决方法：解压高版本的opatch安装包后覆盖原opatch目录

2.备库alert.log报警提示无法找到控制文件自动备份路径

出现原因：RAC主库之前部署过自动RMAN备份脚本，指定了控制文件自动备份路径，但备库并没有此路径

解决方法：进入RMAN，修改该项参数为备库存在的目录

3.参数设置错误而引起GAP,导致自动备份脚本停止运行

出现原因：之前在设置参数时，把主库的log_archive_dest_1参数设置了本地路径归档，如：
alter system set log_archive_dest_1=&＃39;LOCATION=C:\archivelog VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=dg&＃39;;

作为RAC，归档路径在本地的话，其他节点就无法读取，发现后重新设置为USE_DB_RECOVERY_FILE_DEST后，那些在本地的归档日志就成为GAP而无法传递到备库

解决方法：手工复制所有提示缺失的xxx归档到指定位置，再手动执行RMAN自动备份脚本

说明：由于RMAN自动备份脚本里配置了冗余7份,而之搭建DG时手动执行了全库备份，这些手动备份也是算在7份冗余之内的，为了不占用正常备份的配额，DG搭建完成后建议物理删除，然后再crossecheck并清理掉

4.主、备库的alert.log经常会出现TNS错误

fatal NI connect error 12547

TNS-12547 TNS : 丢失连接
ns secondary err code : 12560
ns main err code : 517

TNS-00517 TNS : 丢失连接
nt secondary err code : 54
nt OS err code : 0

出现原因：节点2没有配置tnsnames.ora，造成thread 2的归档日志无法传递到备库，同时也会造成主库日志能传递过去，但无法应用。

解决方法: 把节点1的tnsnames.ora直接复制一个到节点2

说明：其实这个也是造成备库应用出现GAP的最大原因，由于节点2日志传递不到备库，虽然之前的几个归档日志序列相应的applied列的属性值都是YES，但是会造成节点1的日志也不应用，哪怕在节点1切了很多次归档，applied列始终会显示NO，但日志都是可以正常传递过去的