# 跳过损坏的数据块
<a name="Appendix.Oracle.CommonDBATasks.SkippingCorruptBlocks"></a>

要在索引和表扫描期间跳过损坏的数据块，请使用 `rdsadmin.rdsadmin_dbms_repair` 程序包。

下面的过程封装了 `sys.dbms_repair.admin_table` 过程的功能，并且不带任何参数：
+ `rdsadmin.rdsadmin_dbms_repair.create_repair_table`
+ `rdsadmin.rdsadmin_dbms_repair.create_orphan_keys_table`
+ `rdsadmin.rdsadmin_dbms_repair.drop_repair_table`
+ `rdsadmin.rdsadmin_dbms_repair.drop_orphan_keys_table`
+ `rdsadmin.rdsadmin_dbms_repair.purge_repair_table`
+ `rdsadmin.rdsadmin_dbms_repair.purge_orphan_keys_table`

下面的过程采用与其 Oracle 数据库 `DBMS_REPAIR` 程序包中的对应项相同的参数。
+ `rdsadmin.rdsadmin_dbms_repair.check_object`
+ `rdsadmin.rdsadmin_dbms_repair.dump_orphan_keys`
+ `rdsadmin.rdsadmin_dbms_repair.fix_corrupt_blocks`
+ `rdsadmin.rdsadmin_dbms_repair.rebuild_freelists`
+ `rdsadmin.rdsadmin_dbms_repair.segment_fix_status`
+ `rdsadmin.rdsadmin_dbms_repair.skip_corrupt_blocks`

有关处理数据库损坏的更多信息，请参阅 Oracle 文档中的 [DBMS\$1REPAIR](https://docs.oracle.com/en/database/oracle/oracle-database/19/arpls/DBMS_REPAIR.html#GUID-B8EC4AB3-4D6A-46C9-857F-4ED53CD9C948)。

**Example 响应损坏的数据块**  
此示例显示了响应损坏的数据块的基本工作流程。您的步骤将取决于数据块损坏的位置和性质。  
在尝试修复损坏的数据块之前，请仔细查看 [DBMS\$1RECH](https://docs.oracle.com/en/database/oracle/oracle-database/19/arpls/DBMS_REPAIR.html#GUID-B8EC4AB3-4D6A-46C9-857F-4ED53CD9C948) 文档。

**在索引和表扫描期间跳过损坏的数据块**

1. 如果还没有修复表，则运行以下程序来创建修复表。

   ```
   EXEC rdsadmin.rdsadmin_dbms_repair.create_repair_table;
   EXEC rdsadmin.rdsadmin_dbms_repair.create_orphan_keys_table;
   ```

1. 运行以下程序以检查现有记录，适当时清除它们。

   ```
   SELECT COUNT(*) FROM SYS.REPAIR_TABLE;
   SELECT COUNT(*) FROM SYS.ORPHAN_KEY_TABLE;
   SELECT COUNT(*) FROM SYS.DBA_REPAIR_TABLE;
   SELECT COUNT(*) FROM SYS.DBA_ORPHAN_KEY_TABLE;
   
   EXEC rdsadmin.rdsadmin_dbms_repair.purge_repair_table;
   EXEC rdsadmin.rdsadmin_dbms_repair.purge_orphan_keys_table;
   ```

1. 运行以下程序检查损坏的数据块。

   ```
   SET SERVEROUTPUT ON
   DECLARE v_num_corrupt INT;
   BEGIN
     v_num_corrupt := 0;
     rdsadmin.rdsadmin_dbms_repair.check_object (
       schema_name => '&corruptionOwner',
       object_name => '&corruptionTable',
       corrupt_count =>  v_num_corrupt
     );
     dbms_output.put_line('number corrupt: '||to_char(v_num_corrupt));
   END;
   /
   
   COL CORRUPT_DESCRIPTION FORMAT a30
   COL REPAIR_DESCRIPTION FORMAT a30
   
   SELECT OBJECT_NAME, BLOCK_ID, CORRUPT_TYPE, MARKED_CORRUPT, 
          CORRUPT_DESCRIPTION, REPAIR_DESCRIPTION 
   FROM   SYS.REPAIR_TABLE;
   
   SELECT SKIP_CORRUPT 
   FROM   DBA_TABLES 
   WHERE  OWNER = '&corruptionOwner'
   AND    TABLE_NAME = '&corruptionTable';
   ```

1. 使用 `skip_corrupt_blocks` 过程为受影响的表启用或禁用损坏跳过。根据具体情况，您可能还需要将数据提取到新表，然后删除包含损坏的数据块的表。

   运行以下程序为受影响的表启用损坏跳过。

   ```
   begin
     rdsadmin.rdsadmin_dbms_repair.skip_corrupt_blocks (
       schema_name => '&corruptionOwner',
       object_name => '&corruptionTable',
       object_type => rdsadmin.rdsadmin_dbms_repair.table_object,
       flags => rdsadmin.rdsadmin_dbms_repair.skip_flag);
   end;
   /
   select skip_corrupt from dba_tables where owner = '&corruptionOwner' and table_name = '&corruptionTable';
   ```

   运行以下程序禁用损坏跳过。

   ```
   begin
     rdsadmin.rdsadmin_dbms_repair.skip_corrupt_blocks (
       schema_name => '&corruptionOwner',
       object_name => '&corruptionTable',
       object_type => rdsadmin.rdsadmin_dbms_repair.table_object,
       flags => rdsadmin.rdsadmin_dbms_repair.noskip_flag);
   end;
   /
   
   select skip_corrupt from dba_tables where owner = '&corruptionOwner' and table_name = '&corruptionTable';
   ```

1. 完成所有修复工作后，运行以下过程删除修复表。

   ```
   EXEC rdsadmin.rdsadmin_dbms_repair.drop_repair_table;
   EXEC rdsadmin.rdsadmin_dbms_repair.drop_orphan_keys_table;
   ```