首页 > 代码库 > 搭建一个Oracle到Oracle的Goldengate双向复制环境
搭建一个Oracle到Oracle的Goldengate双向复制环境
目标：搭建一个Oracle到Oracle的Goldengate双向复制环境（支持ＤＤＬ＋ＤＭＬ）。
环境：
ＯＳ：Red Hat Enterprise Linux Server release 5.5 (Tikanga)
ＤＢ：Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
gg1和gg2互为source和target，所处ＯＳ和所安ＤＢ均一样。
一、安装ＧＧ
在gg1和gg2都执行如下操作：
（１）添加环境变量，在/home/oracle/.bash_profile文件里添加如下内容：
export PATH=/u01/ggate:$PATH
exportLD_LIBRARY_PATH=/u01/ggate:$LD_LIBRARY_PATH
export ggtest=/u01/ggate
这里的GG 和Oracle 使用的是相同的用户，所以把GG 的变量加上就可以了。加载刚刚设置的环境变量：
source/home/oracle/.bash_profile
（２）使用ggsci工具，创建必要的目录
gg1:/u01/ggate> ggsci
--调用ggsci 工具
Oracle GoldenGate Command Interpreter forOracle
Version 11.1.1.1OGGCORE_11.1.1_PLATFORMS_110421.2040
Linux, x64, 64bit (optimized), Oracle 11gon Apr 21 2011 22:42:14
Copyright (C) 1995, 2011, Oracle and/or itsaffiliates. All rights reserved.
GGSCI (gg1) 1> create subdirs
--使用ggsci 工具创建目录
Creating subdirectories under currentdirectory /u01/ggate
Parameter files /u01/ggate/dirprm: created
Report files /u01/ggate/dirrpt: created
Checkpoint files /u01/ggate/dirchk: created
Process status files /u01/ggate/dirpcs: created
SQL script files /u01/ggate/dirsql: created
Database definitions files /u01/ggate/dirdef: created
Extract data files /u01/ggate/dirdat: created
Temporary files /u01/ggate/dirtmp: created
Veridata files /u01/ggate/dirver: created
Veridata Lock files /u01/ggate/dirver/lock: created
Veridata Out-Of-Sync files /u01/ggate/dirver/oos: created
Veridata Out-Of-Sync XML files/u01/ggate/dirver/oosxml: created
Veridata Parameter files /u01/ggate/dirver/params: created
Veridata Report files /u01/ggate/dirver/report: created
Veridata Status files /u01/ggate/dirver/status: created
Veridata Trace files /u01/ggate/dirver/trace: created
Stdout files /u01/ggate/dirout: created
GGSCI (gg1) 2>
以上就是GG 的安装，在source 和target database 都执行。
二、分别配置ｓｏｕｒｃｅ和ｔａｒｇｅｔ
（１）GoldenGate通过抓取源端数据库重做日志进行分析，将获取的数据应用到目标端，实现数据同步。因此，源数据库需要必须处于归档模式，并启用附加日志和强制日志。
归档模式、附加日志、强制日志
--查看
SQL> select log_mode,supplemental_log_data_min,force_logging from v$database;
LOG_MODE SUPPLEME FOR
------------ -------- ---
ARCHIVELOG NO NO
--修改
（1）archivelog
SQL>shutdown immediate
SQL>startup mount
SQL>alter database archivelog;
SQL>alter database open;
（2）force logging
SQL>alterdatabase force logging;
（3）supplemental log data
SQL>alterdatabase add supplemental log data;
（２）禁用Recycle Bin
如果启用DDL 支持，必须关闭recycle bin。官网的解释如下：
If the recyclebin is enabled, the Oracle GoldenGate DDL trigger session receives implicitrecycle bin DDL operations that cause the trigger to fail.
Oracle 11g：
SQL> alter system set recyclebin=offscope=spfile;
System altered.
如果数据库是10g，需要关闭recyclebin并重启；或者手工purge recyclebin。
（３）创建存放DDL 信息的user并赋权
SQL> create user ggtest identified by oracle default tablespace users temporary tablespace temp;
User created.
SQL> grant connect,resource to ggtest;
Grant succeeded.
SQL> grant execute on utl_file to ggtest;
Grant succeeded.
退出所有使用Oracle 的session，然后使用SYSDBA权限的用户执行如下脚本：
gg1:/u01/ggate> echo $ggate
/u01/ggate
--进入GG的目录，然后调用脚本：
gg1:/home/oracle> cd $ggate
gg1:/u01/ggate> sqlplus / as sysdba;
SQL*Plus: Release 11.2.0.3.0 Production onTue Nov 8 19:41:58 2011
Copyright (c) 1982, 2011, Oracle. All rights reserved.
Connected to:
Oracle Database 11g Enterprise EditionRelease 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Miningand Real Application Testing options
--脚本1：
SQL> @marker_setup.sql;
Marker setup script
You will be prompted for the name of aschema for the GoldenGate database objects.
NOTE: The schema must be created prior torunning this script.
NOTE: Stop all DDL replication beforestarting this installation.
--输入我们之前创建的用户名：
Enter GoldenGate schema name:ggtest
Marker setup table script complete, runningverification script...
Please enter the name of a schema for theGoldenGate database objects:
Setting schema name to ggtest
MARKER TABLE
-------------------------------
OK
MARKER SEQUENCE
-------------------------------
OK
Script complete.
--脚本2：
SQL> @ddl_setup.sql;
GoldenGate DDL Replication setup script
Verifying that current user has privilegesto install DDL Replication...
You will be prompted for the name of aschema for the GoldenGate database objects.
NOTE: For an Oracle 10g source, the systemrecycle bin must be disabled. For Oracle 11g and later, it can be enabled.
--注意这里提示我们在10g里，必须关闭recycle bin，在11g以后的版本，可以不用关闭。
NOTE: The schema must be created prior torunning this script.
NOTE: Stop all DDL replication beforestarting this installation.
--提示输入GG的用户：
Enter GoldenGate schema name:ggtest
You will be prompted for the mode ofinstallation.
To install or reinstall DDL replication,enter INITIALSETUP
To upgrade DDL replication, enter NORMAL
--这里让我们选择安装模式： install 和 reinstall 选择INITIALSETUP
Enter mode of installation:INITIALSETUP
Working, please wait ...
Spooling to file ddl_setup_spool.txt
Checking for sessions that are holdinglocks on Oracle Golden Gate metadata tables ...
Check complete.
Using ggtest as a GoldenGate schema name,INITIALSETUP as a mode of installation.
Working, please wait ...
DDL replication setup script complete,running verification script...
Please enter the name of a schema for theGoldenGate database objects:
Setting schema name to ggtest
DDLORA_GETTABLESPACESIZE STATUS:
……
STATUS OF DDL REPLICATION
-------------------------------------------------------------------------------------------------------
SUCCESSFUL installation of DDL Replicationsoftware components
Script complete.
--脚本3：
SQL> @role_setup.sql;
GGS Role setup script
This script will drop and recreate the roleGGS_GGSUSER_ROLE
To use a different role name, quit thisscript and then edit the params.sql script to change the gg_role parameter tothe preferred name. (Do not run the script.)
You will be prompted for the name of aschema for the GoldenGate database objects.
NOTE: The schema must be created prior torunning this script.
NOTE: Stop all DDL replication beforestarting this installation.
--同样输入GG用户名：
Enter GoldenGate schema name:ggtest
Wrote file role_setup_set.txt
PL/SQL procedure successfully completed.
Role setup script complete
Grant this role to each user assigned tothe Extract, GGSCI, and Manager processes, by using the following SQL command:
--这里提示我们赋权给相关的用户：
GRANT GGS_GGSUSER_ROLE TO<loggedUser>
where <loggedUser> is the userassigned to the GoldenGate processes.
--脚本4：赋权
SQL> grant GGS_GGSUSER_ROLE to ggtest;
Grant succeeded.
--脚本5：
SQL> @ddl_enable.sql;
Trigger altered.
注意这里脚本创建的table都是使用默认的名称，当然也可以修改这些table的默认名。
三、测试
注意：
（1） 目标库的用户名和对象名称可以与源端不同，关键在于配置文件中要能够正确匹配。
（2） 配置源和目标两端tnsnames，保持互联互通。（记得启动lsnrctl）
gg1:
SQL> create user ssgg identified by oracle default tablespace users temporary tablespace temp;
User created.
SQL> grant connect,resource,dba to ssgg;
Grant succeeded.
gg2:
SQL> create user ttgg identified by oracle default tablespace users temporary
tablespace temp;
User created.
SQL> grant connect,resource,dba to ttgg;
Grant succeeded.
参数设置
配置gg1，添加checkpoint表：
GGSCI (gg1) 1> dblogin userid ggtest,password oracle
Successfully logged into database.
GGSCI (gg1) 2> view params ./GLOBAL
ggschema ggtest
checkpointtable ggtest.checkpoint
GGSCI (gg1) 27> view params mgr
PORT 7809 --指定管理端口
dynamicportlist 7810-7900 --动态端口列表，当指定端口不可用时，管理进程会自动选择一个可用端口。最大可指定256个端口
autostart er * --mgr开启时自动开启所有的extract and replicat process
autorestart er *,retries 5,waitminutes 2 --extract and replicat processes fail mgr automatically start
参数每2分钟重启一次，重启5次后放弃
1、 t o c o n f i g u r e t h e p r i m a r y e x t r a c t g o u p
GGSCI (gg1) 10> add extract ext01,tranlog,begin now
EXTRACT added.
GGSCI (gg1) 13> add exttrail /u01/ggate/dirdat/sd,extract ext01
EXTTRAIL added.
GGSCI (gg1) 28> view params ext01
extract ext01 --提取进程组的名字
userid ggtest, password oracle --同步用户
exttrail /u01/ggate/dirdat/sd --trail 文件目录和标识（类似于sd*这样文件)，命名只能是2个字符。
tranlogoptions excludeuser ggtest --避免循环复制的出现
DDL INCLUDE ALL
DDLOPTIONS ADDTRANDATA
table ssgg.*; --同步的表或用户下所有的表
当你需要同步除Goldengate和Oracle数据库自带的用户外所有用户的DDL操作时，可在源端主抽取进程中加入下面的内容
ddl include all
然后再在目标端Replicat进程参数文件中加入DDL错误的处理语句即可。
2、 t o c o n f i g u r e t h e d a t a p u m p
GGSCI (gg1) 19> add extract p01,exttrailsource /u01/ggate/dirdat/sd,begin now
EXTRACT added.
GGSCI (gg1) 20> add rmttrail /u01/ggate/dirdat/pd,extract p01
RMTTRAIL added.
GGSCI (gg1) 29> view params p01
extract p01
userid ggtest,password oracle
rmthost 192.168.50.231,mgrport 7809 --指定远端IP和MGR端口
rmttrail /u01/ggate/dirdat/pd
table ssgg.*;
3、 t o c o n f i g u r e t h e r e p l i c a t g r o u p
GGSCI (gg1) 3> add replicat r02,exttrail /u01/ggate/dirdat/dp,begin now
ERROR: No checkpoint table specified for ADD REPLICAT.
GGSCI (gg1) 4> add replicat r02,exttrail /u01/ggate/dirdat/dp,checkpointtable ggtest.checkpoint
REPLICAT added.
GGSCI (gg1) 30> view params r02
replicat r02
assumetargetdefs --定义目标表和源表使用MAP语法有相同的列结构，当出现热点时，不查看源结构而直接从源定义文件查
看
userid ggtest,password oracle
DISCARDFILE /u01/ggate/dirdat/sdisc.dsc,append,megabytes 100
DDL INCLUDE MAPPED
DDLERROR DEFAULT IGNORE RETRYOP
map ttgg.*, target ssgg.*;
启动所有进程：
GGSCI (gg1) 3> start mgr
Manager started.
GGSCI (gg1) 4> info all
Program Status Group Lag at Chkpt Time Since Chkpt
MANAGER RUNNING
EXTRACT RUNNING EXT01 00:00:00 00:00:07
EXTRACT RUNNING P01 00:00:00 00:00:01
REPLICAT RUNNING R02 00:00:00 00:00:08
配置gg2，添加checkpoint表：
GGSCI (gg2) 1> dblogin userid ggtest,password oracle
Successfully logged into database.
GGSCI (gg2) 25> view params ./GLOBAL
ggschema ggtest
checkpointtable ggtest.checkpoint
GGSCI (gg2) 29> view params mgr
PORT 7809
dynamicportlist 7810-7900
autostart er *
autorestart er *,retries 5,waitminutes 2
c o n f i g u r a t i o n f r o m s e c o u n d a r y s y s t e m t o p r i m a r y s y s t e m
1 、t o c o n f i g u r e t h e p r i m a r y e x t r a c t g r o u p
GGSCI (gg2) 12> add extract ext02,tranlog,begin now
EXTRACT added.
GGSCI (gg2) 13> add exttrail /u01/ggate/dirdat/td,extract ext02
EXTTRAIL added.
GGSCI (gg2) 26> view params ext02
extract ext02
userid ggtest, password oracle
exttrail /u01/ggate/dirdat/td
tranlogoptions excludeuser ggtest
DDL INCLUDE ALL
DDLOPTIONS ADDTRANDATA
table ttgg.*;
2 、t o c o n f i g u r e t h e p u m p
GGSCI (gg2) 17> add extract p02,exttrailsource /u01/ggate/dirdat/td,begin now
EXTRACT added.
GGSCI (gg2) 19> add rmttrail /u01/ggate/dirdat/dp,extract p02
RMTTRAIL added.
GGSCI (gg2) 27> view params p02
extract p02
userid ggtest,password oracle
rmthost 192.168.50.230,mgrport 7809
rmttrail /u01/ggate/dirdat/dp
table ttgg.*;
3.t o c o n f i g u r e t h e r e p l i c a t g r o u p
GGSCI (gg2) 7> add replicat r01,exttrail /u01/ggate/dirdat/pd,begin now
ERROR: No checkpoint table specified for ADD REPLICAT.
GGSCI (gg2) 8> add replicat r01,exttrail /u01/ggate/dirdat/pd,checkpointtable ggtest.checkpoint;
REPLICAT added.
GGSCI (gg2) 28> view params r01
replicat r01
assumetargetdefs
userid ggtest,password oracle
DISCARDFILE /u01/ggate/dirdat/sdisc.dsc,append,megabytes 100
DISCARDFILE不能直接在安装目录/u01/ggate/，进程可能起不来。
DDL INCLUDE MAPPED
DDLERROR DEFAULT IGNORE RETRYOP
map ssgg.*, target ttgg.*;
启动所有进程：
GGSCI (gg2) 3> start mgr
Manager started.
GGSCI (gg2) 4> info all
Program Status Group Lag at Chkpt Time Since Chkpt
MANAGER RUNNING
EXTRACT RUNNING EXT02 00:00:00 00:00:03
EXTRACT RUNNING P02 00:00:00 00:00:07
REPLICAT RUNNING R01 00:00:14 00:00:04
数据同步测试：
Gg1：
SQL> conn ssgg/oracle
Connected.
SQL> select * from tab;
TNAME TABTYPE CLUSTERID
------------------------------ ------- ----------
SSGG TABLE
TTGG TABLE
SQL> create table test as select * from ssgg;
Table created.
SQL> select * from test;
ID
----------
1
5
SQL> commit;
Commit complete.
然后查看gg2，看数据是否同步
Gg2：
SQL> conn ttgg/oracle
Connected.
SQL> select * from tab;
TNAME TABTYPE CLUSTERID
------------------------------ ------- ----------
SSGG TABLE
TTGG TABLE
SQL> /
TNAME TABTYPE CLUSTERID
------------------------------ ------- ----------
SSGG TABLE
TEST TABLE
TTGG TABLE
SQL> insert into test values(10);
1 row created.
SQL> commit;
Commit complete.
SQL> select * from test;
ID
----------
1
5
10
再返回gg1，查看数据是否同步：
SQL> host hostname
gg1
SQL> select * from test;
ID
----------
1
5
10
至此，Oracle到Oracle的双向数据同步成功(DML+DDL)。
实现双向复制，一般用于双业务中心环境下在实际应用中。双向复制（ＤＤＬ＋ＤＭＬ）面临着许多问题，主要有如下几点：
1. 如果两个库同时更新同一条记录如何处理？调整业务规则
2. 如果网络出现失败如何处理？支持断点续传
3. 如果数据不同步后如何修复？导出源端，导入目标端
总结：
双向复制避免数据冲突是难点，解决这个难点根本还是调整业务：
a) 只在一段开展业务
b) 两端开展不同业务，涉及不同数据集
c) 两端开展相同业务，但依据地域或其他条件对数据予以区分，两边不操作同一条数据
补充：
在oracle中可以通过rowid来定位某条记录，但是目标端的数据库和源端数据库的数据库可能完全不一样，所以无法通过rowid来确定源端数据库的逻辑变化，这时附加日志supplemental log便登上了表演的舞台。数据库在开启附加日志功能后，对于源端的修改操作，oracle会同时追加能够唯一标示记录的列到redo log。这样目标端数据库就可以知道源端发生了哪些具体的变化。
在目标库上创建checkpointtable ，此表中保存的检查点指示 Replicat 进程的当前读写位置。这用于在进程需要重启时或者在服务器出现任何故障或网络停顿时防止数据丢失。如果没有这一机制，将导致数据丢失。另一个好处是，通过使用检查点，多个 Extract 或 Replicat 进程可通过相同的一组线索进行读取。
对于以批处理模式运行的 Extract 和 Replicat 进程，检查点不是必需的，因为这样的进程总是可以重启。然而，在 Extract 和 Replicat 进程连续工作的情况下，检查点是必需的。检查点通常保存在 dirchk 子目录的文件中，但是对于 Replicat，可以选择在数据库的检查点表中保存检查点。
搭建一个Oracle到Oracle的Goldengate双向复制环境