Transporting Tablespace Between Database

  • 【测试目的】

    利用Oracle TTS(transport tablespace)特性实现表空间合并

  • 【主要步骤】

    1. 确定源平台/目的平台信息以及以及字节序信息

    2. 确定待传输的表空间是否独立 (self-contained)

    3. 生成transportable tablespace set

      transportable tablespace set包含待传输表空间所有数据文件以及待传输表空间结构信息(metadata)导出文件;

      如果源平台/目的平台的字节序信息不同, 就必须将tablespace set的字节序信息转换为目的平台;

      既可以在源平台上转换, 也可以在目的平台转换;

    4. 传输(scp/rsync) tablespace set

    5. 导入tablespace set

  • 【测试一】

    从DB01将user01用户相关的表空间传输到DB02

    实例信息:

    user01DB011.1.1.11.1.1.2
    user02DB022.2.2.22.2.2.3

    数据文件:

      SQL> select file_name,tablespace_name from dba_data_files where tablespace_name in ('user01_DAT','user01_IDX');
    
      FILE_NAME                                          TABLESPACE_NAME
    -------------------------------------------------- ------------------------------
    /data1/oradata/DB01/r221dat009.dbf user01_DAT
    /data1/oradata/DB01/r221dat010.dbf user01_DAT ......
    /data1/oradata/DB01/r221idx007.dbf user01_IDX
    /data1/oradata/DB01/r221idx008.dbf user01_IDX 30 rows selected.

    测试示例:

    • Step 1: 确定平台信息以及字节信息

    source platform:

      SQL> SELECT d.PLATFORM_NAME, ENDIAN_FORMAT FROM V$TRANSPORTABLE_PLATFORM tp, V$DATABASE d WHERE tp.PLATFORM_NAME = d.PLATFORM_NAME;
    
      PLATFORM_NAME      ENDIAN_FORMAT
    -------------------------- --------------
    Linux x86 64-bit Little

    target platform:

      SQL> SELECT d.PLATFORM_NAME, ENDIAN_FORMAT FROM V$TRANSPORTABLE_PLATFORM tp, V$DATABASE d WHERE tp.PLATFORM_NAME = d.PLATFORM_NAME;
    
      PLATFORM_NAME      ENDIAN_FORMAT
    -------------------------- --------------
    Linux x86 64-bit Little

    从查询结果看到: 平台和字节序信息相同, 因此不需要进行convert操作

    • Step 2: 确认Self-Contained Set of Tablespaces

    执行存储过程:

    DBMS_TTS.TRANSPORT_SET_CHECK检查表空间是否独立

      SQL>EXECUTE DBMS_TTS.TRANSPORT_SET_CHECK('user01_DAT,user01_IDX',TRUE);
    
      PL/SQL procedure successfully completed.
    
      SQL> select * from TRANSPORT_SET_VIOLATIONS;
    
      no rows selected
    • Step 3: 生成Transport Tablespace Set

      1. 将要传输的相关表空间read-only

         SQL> alter tablespace user01_DAT read only;
        SQL> alter tablespace user01_IDX read only;
      2. 导出metadata信息

        2.1创建导出目录

         SQL> create directory dump_dir as '/u/dump/';
        SQL> grant read,write on directory dump_dir to PUBLIC;

        2.2调用expdp工具

         expdp dumpfile=expdat.dmp directory=dump_dir transport_tablespaces = user01_DAT,user01_IDX transport_full_check=y
        
         Export: Release 11.2.0.2.0 - Production on Thu Sep 26 11:45:33 2013
        
         Copyright (c) 1982, 2009, Oracle and/or its affiliates.  All rights reserved.
        
         Username:******
        
         Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.2.0 - 64bit Production
        With the Partitioning, OLAP, Data Mining and Real Application Testing options
        Starting "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01": system/******** dumpfile=expdat.dmp directory=dump_dir transport_tablespaces=user01_DAT,user01_IDX transport_full_check=y
        Processing object type TRANSPORTABLE_EXPORT/PLUGTS_BLK
        ORA-39126: Worker unexpected fatal error in KUPW$WORKER.UNLOAD_METADATA [PLUGTS_BLK]
        ORA-00904: "Q"."TABOBJ_NUM": invalid identifier ORA-06512: at "SYS.DBMS_SYS_ERROR", line 95
        ORA-06512: at "SYS.KUPW$WORKER", line 8358 ----- PL/SQL Call Stack -----
        object line object
        handle number name
        0x5ba4f3e20 19208 package body SYS.KUPW$WORKER
        0x5ba4f3e20 8385 package body SYS.KUPW$WORKER
        0x5ba4f3e20 2935 package body SYS.KUPW$WORKER
        0x5ba4f3e20 9054 package body SYS.KUPW$WORKER
        0x5ba4f3e20 1688 package body SYS.KUPW$WORKER
        0x5ba4f7e60 2 anonymous block
        Job "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01" stopped due to fatal error at 11:13:56

        在这里调用expdp工具时, 遇到错误:

        encounter some error: ORA-39126: Worker unexpected fatal error in KUPW$WORKER.UNLOAD_METADATA [PLUGTS_BLK]

        for more information: http://martincarstenbach.wordpress.com/2011/05/10/slight-11-2-0-2-caveat-with-tts-and-missing-xmldb/

        解决办法:

        A. 使用exp/imp工具

         	exp file=/u/dump/expdat.dmp log=/u/dump/expdat.log transport_tablespace=y tablespaces=user01_DAT,user01_IDX

        B. 编译后,使用expdp/impdp工具

         	conn / as sysdba
        @?/rdbms/admin/catbundle.sql psu apply
        @?/rdbms/admin/catqm.sql xdb sysaux temp yes
        @?/rdbms/admin/utlrp.sql expdp dumpfile=expdat.dmp directory=dump_dir transport_tablespaces = user01_DAT,user01_IDX transport_full_check=y
      3. 如果Step 1中的平台信息, 以及字节序信息不同, 那么就需要对表空间进行convert; 否则直接跳过, 进行Step 4操作

        这里假设target platform是Microsoft Windows NT平台, 需要在source platform上convert数据文件到临时目录:

         RMAN TARGET / NOCATALOG
        RMAN> convert tablespace user01_DAT,user01_IDX to platform 'Microsoft Windows NT' format '/temp/%U';
    • Step 4: 传输Tablespace Set

        rsync -a --progress  [email protected]:/data1/oradata/DB01/r221*dbf /data1/oradata/DB01/
      rsync -a --progress [email protected]:/u/dump/expdat.* /u/

      如果是将表空间传输到不同平台不同字节序的目的端, 并且Step 3中没有做convert操作, 然而不管数据文件是否被convert过, 都必须移动到特定的目的端位置;

      使用RMAN命令convert传输过来的数据文件为目的端字节序信息, 并且将结果存储在/data1/oradata/DB01/

      这里假设数据文件存放在目的端目录:/data1/dbbak/test/,target platform是Microsoft Windows NT平台:

        RMAN> convert datafile
      2> '/data1/dbbak/test/r231dat005.dbf'
      ......
      13> '/data1/dbbak/test/r231idx004.dbf'
      14> to platform="Microsoft Windows NT"
      15> from platform="Linux x86 64-bit"
      16> db_file_name_convert=
      17> "/data1/dbbak/test/","/data1/oradata/DB01/"
      18> parallelism=5;

      备注:

      1. 省略号出需要填写全部的文件信息;
      2. 在target platform上RMAN是通过文件名来辨识数据文件的, 而不是表空间名称. 因为目的端是无法知道表空间名称的, 直到metadata导入到实例;
      3. 参数: to platform和from platform信息是可选的, RMAN通过存在的数据文件确定source platform信息, 以及运行convert主机上默认的target platform信息;
    • Step 5: Import the Tablespace Set

      1. 在目标平台上, 导入metadata信息之前用户必须存在

         SQL> create user user01 identified by user01;
        SQL> grant connect to user01;

        或者

        可以用下面sql迁移source上的用户和权限

        1.1 创建用户

         SQL> select 'create user '||a.name||' identified by values "'||a.password||'";' from user$ a,dba_users b where a.user#=b.user_id and b.account_status='OPEN' and a.name not in ('SYS','SYSTEM','PERFSTAT');

        1.2 设置用户默认的表空间

         SQL> select 'alter user '||a.name||' default tablespace '||b.default_tablespace||';' from user$ a,dba_users b where a.user#=b.user_id and b.account_status='OPEN' and a.name not in ('SYS','SYSTEM','PERFSTAT');

        1.3 初始化用户权限

         SQL> select 'grant create session to '||a.name||';' from user$ a,dba_users b where a.user#=b.user_id and b.account_status='OPEN' and a.name not in ('SYS','SYSTEM','PERFSTAT');
        SQL> select 'grant '||c.PRIVILEGE||' to '||d.name||';' from (select distinct a.name from user$ a,dba_users b,dba_sys_privs c where a.user#=b.user_id and a.name=c.grantee and b.account_status='OPEN' and a.name not in ('SYS','SYSTEM','PERFSTAT')) d,dba_sys_privs c where d.name=c.grantee;
      2. Import metadata信息

        2.1 使用imp/exp工具

         imp parfile=/u/par.f

        --参数:parfile可以用来指定impdp参数

        /u/par.f内容:

         file=/u/expdat.dmp
        log=/u/impdat.log
        transport_tablespace=y
        tablespace=user01_DAT,user01_IDX
        datafiles=
        /data1/oradata/DB01/r221dat015.dbf,
        ......
        /data1/oradata/DB01/r221idx002.dbf,
        /data1/oradata/DB01/r221idx001.dbf

        备注:

        1. 省略号处需要列举所有数据文件信息;

        2.2 使用expdp/impdp工具

         impdp parfile=/u/par.f

        --参数: parfile可以用来指定impdp参数

        /u/par.f内容:

         dumpfile=expdat.dmp
        directory=dump_dir
        transport_datafiles=
        /data1/oradata/DB02/r231dat005.dbf,
        ......
        /data1/oradata/DB02/r231idx003.dbf,
        /data1/oradata/DB02/r231idx004.dbf
        logfile=tts_impdp.log

        备注:

        1. 省略号处需要列举所有数据文件信息;

--END--

05-27 13:13