SQL load direct path load index 无效的原因

Index (unique) state changing to unusable status after data loading using SQL*Loader. The steps below are executed to load the data:

1/ disable constraint--如果不是单独建的index, 对应index会drop掉的

2/ load data using SQL*Loader

3/ remove duplicate records

4/ enable constraint

The execution of step 3 fails due to an index in unusable state:

\n removing duplicate records from the table ... \n

DELETE FROM <TABLE_NAME1> WHERE rowid not in (SELECT MIN(rowid) FROM <TABLE_NAME1> GROUP BY <COLUMN_NAME>)---rowid在期间不能有table move 操作

*

ERROR at line 1:

ORA-01502: index '<INDEX_NAME>_PK' or partition of such index is

in unusable state

Checking the available indexes and their status from dba_indexes, we have:

|-------------------|------------|----------|-----------------|
| INDEX_NAME | INDEX_TYPE | STATUS | TABLE_NAME |
| <INDEX_NAME>_PK | NORMAL | UNUSABLE | <TABLE_NAME1> |
| <INDEX_NAME>_PK | NORMAL | VALID | <TABLE_NAME2> |

CHANGES

CAUSE

Some integrity constraints are automatically disabled During a direct path load. The constraints that remain enabled during a direct path load are:

  • NOT NULL
  • UNIQUE
  • PRIMARY KEY (unique-constraints on not-null columns)--PK自动not null

这么说不就是外键 check disable了

  • NOT NULL constraints are checked at column array build time. Any row that violates the NOT NULL constraint is rejected.

Even though UNIQUE constraints remain enabled during direct path loads, any row that violates those constraints is loaded anyway (this is different than in conventional path in which such rows would be rejected).

When indexes are rebuilt at the end of the direct path load, UNIQUE constraints are verified and if a violation is detected, then the index will be left in an unusable state.

不拒绝

Indexes Left in an Unusable State


SQL*Loader leaves indexes in an Index Unusable state when the data segment being loaded becomes more up-to-date than the index segments that index it.

Any SQL statement that tries to use an index that is in an Index Unusable state returns an error. The following conditions cause a direct path load to leave an index or a partition of a partitioned index in an Index Unusable state:

  • SQL*Loader runs out of space for the index and cannot update the index.

  • The data is not in the order specified by the SORTED INDEXES clause.

  • There is an instance failure, or the Oracle shadow process fails while building the index.

  • There are duplicate keys in a unique index.

  • Data savepoints are being used, and the load fails or is terminated by a keyboard interrupt after a data savepoint occurred.

Hence this is a expected behavior as:

  • Some duplicate rows are getting loaded and indexes are going to unusable state when unique indexes are used.

  • Or -

  • SQL*Loader leaves indexes in an Index Unusable state when the data segment being loaded becomes more up-to-date than the index segments that index it.

SOLUTION

To overcome the issue:

  • In few cases using conventional path load has resolved the issue.
  • Or -
  • Using normal index instead of unique index will not affect the index status.
  • 先建立普通index 加unique key 控制
  • Or -
  • One can follow below workaround for unique indexes.

  • 本来是个死局,重复的删不了,index不能rebuild,只有drop,

    • Drop the offending index and load the data.
    • Detect duplicate rows
    • Remove the duplicate rows and recreate the index.
相关推荐
RestCloud17 小时前
SQL Server到Hive:批处理ETL性能提升30%的实战经验
数据库·api
RestCloud17 小时前
为什么说零代码 ETL 是未来趋势?
数据库·api
ClouGence19 小时前
CloudCanal + Paimon + SelectDB 从 0 到 1 构建实时湖仓
数据库
DemonAvenger1 天前
NoSQL与MySQL混合架构设计:从入门到实战的最佳实践
数据库·mysql·性能优化
AAA修煤气灶刘哥2 天前
后端人速藏!数据库PD建模避坑指南
数据库·后端·mysql
RestCloud2 天前
揭秘 CDC 技术:让数据库同步快人一步
数据库·api
得物技术2 天前
MySQL单表为何别超2000万行?揭秘B+树与16KB页的生死博弈|得物技术
数据库·后端·mysql
可涵不会debug2 天前
【IoTDB】时序数据库选型指南:工业大数据场景下的技术突围
数据库·时序数据库
ByteBlossom2 天前
MySQL 面试场景题之如何处理 BLOB 和CLOB 数据类型?
数据库·mysql·面试