操作分区表对global和local索引的影响-CSDN博客

使用分区表，要非常注意索引。因为删除、move分区都会造成索引失效。小表的索引和非主键、唯一建索引还好，重建即可。

如果是几十G的表，重建索引花费巨大，会严重影响查询性能;
如果是主键或者唯一键索引，索引失效会插入重复的数据，带来的麻烦更大。

所以，我们一定要重视分区表操作，避免引起索引失效。

哪些操作会引起索引失效？
ADD PARTITION | SUBPARTITION (hash)
COALESCE PARTITION | SUBPARTITION (hash)
DROP PARTITION | SUBPARTITION
EXCHANGE PARTITION | SUBPARTITION
MERGE PARTITION | SUBPARTITION
MOVE PARTITION | SUBPARTITION
SPLIT PARTITION | SUBPARTITION
TRUNCATE PARTITION | SUBPARTITION

上面所有操作都会引起global索引失效;

其中下面的操作会引起local索引失效。
MERGE PARTITION
MOVE PARTITION
SPLIT PARTITION
COALESCE PARTITION
EXCHANGE PARTITION

怎么解决索引失效问题，那就是加上update indexes
加上update indexes，以上任何操作不会引起glocal索引失效;
加上update indexes，以上操作中除了EXCHANGE PARTITION操作以外，不会引起local 索引失效。
*EXCHANGE PARTITION操作是个很特殊的操作，加上update indexes参数，EXCHANGE PARTITION依然会造成local 索引失效。
需要注意的是，如果分区中不含数据，上面的操作都不会引起索引失效（EXCHANGE PARTITION除外）。

下面是实验过程：
注意：
本实验都是range分区
准备基表,id列作为分区key
create table part_test(id number,name varchar2(10))
partition by range (id)
(
partition p1 values less than (5),
partition p2 values less than (10),
partition p3 values less than (15)
);

SQL> select table_name,partition_name,HIGH_VALUE from dba_tab_partitions where table_name='PART_TEST' order by table_name,partition_name;

TABLE_NAME PARTITION_NAME HIGH_VALUE
------------------------------ ------------------------------ --------------------------------------------------------------------------------
PART_TEST P1 5
PART_TEST P2 10
PART_TEST P3 15

一、准备基表及索引

global和local信息不在同一个数据字典中;
global索引信息==>dba_indexes
local索引信息==>dba_ind_partitions/user_ind_partitions
*可以通过dba_indexes判断索引类型，如果status等于VALID或者UNUSABLE，索引类型是global;
status等于N/A，索引则是local索引。

1.建索引
创建global索引
create index ind_id on part_test(id);

创建local索引
create index ind_name on part_test(name) local;

PART_TEST上两个索引名、表名、及索引状态值
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where table_name='PART_TEST';

INDEX_NAME TABLE_OWNER STATUS

------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID --global index
IND_NAME DOWNLOAD N/A --local index

2.索引状态值
查看global索引ind_id 信息
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

查看local索引IND_NAME信息
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE

IND_NAME P2 10 USABLE
IND_NAME P3 15 USABLE

二、 add partition

1.not plus "update indexes"
SQL> alter table part_test add partition p4 values less than (20);

Table altered.

索引都可用
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS

------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P2 10 USABLE
IND_NAME P3 15 USABLE
IND_NAME P4 20

2. plus "update indexes"
SQL> alter table part_test add partition p5 values less than (25) update indexes;

Table altered.

索引同样都可用
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P2 10 USABLE

IND_NAME P3 15 USABLE
IND_NAME P4 20 USABLE
IND_NAME P5 25 USABLE

结论：add partition时，不加update indexes不会破坏local及global索引的可用性;当然，加上update indexes更好更保险。

三、 drop partition

1.删除没有数据的分区
<1>not plus "update indexes" ，同时分区中没有任何数据
SQL> alter table part_test drop partition p4;

Table altered.

删除分区后，索引依然可用。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS

------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P2 10 USABLE

IND_NAME P3 15 USABLE
IND_NAME P5 25 USABLE

<2>plus "update indexes" ，同时分区中没有任何数据
SQL> alter table part_test drop partition p5 update indexes;

Table altered.

删除分区后，索引依然可用。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS

------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P2 10 USABLE
IND_NAME P3 15 USABLE

小结：分区中没有数据，不加update index ，drop该索引也不会引起local及global索引失效。

2.删除含有数据的分区
insert into part_test values(1,'tom'); --落在p1分区
insert into part_test values(6,'lucy'); --落在p2分区
insert into part_test values(11,'lucy'); --落在p3分区

SQL> select * from part_test partition(p1);

ID NAME

---------- ----------
1 tom

SQL>
SQL> select * from part_test partition(p2);

ID NAME
---------- ----------
6 lucy

SQL> select * from part_test partition(p3);

ID NAME
---------- ----------
11 lucy

<1>not plus "update indexes"
SQL> alter table part_test drop partition p3;

Table altered.

global索引失效
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD UNUSABLE

local索引正常
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE

IND_NAME P2 10 USABLE

小结：分区中含有数据，删除分区会造成global索引失效;而local 索引正常。

<2>plus "update indexes"
首先修复global索引IND_ID,通过rebuild
alter index IND_ID rebuild;

再添加几个实验分区
alter table part_test add partition p2 values less than(10);
alter table part_test add partition p3 values less than(15);
alter table part_test add partition p4 values less than(20);

insert into part_test values(6,'lucy'); --落在p2分区
insert into part_test values(11,'lucy'); --落在p3分区

SQL> alter table part_test drop partition p2 update indexes;

Table altered.

SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------

IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE
IND_NAME P4 20 USABLE

小结：分区中含有数据，不加update indexes，会造成global索引失效，而local索引不受影响。

四、 move partition
1.分区含有数据，不加update indexes
SQL> alter table part_test move partition p3;

Table altered.

global索引失效
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS

------------------------------ ------------------------------ --------
IND_ID DOWNLOAD UNUSABLE

分区对应的local索引失效，而其他分区的local索引不受影响。
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 UNUSABLE

IND_NAME P4 20 USABLE

恢复失效索引
global 索引
SQL> alter index IND_ID rebuild;

Index altered.

local索引:rebuild 失效的lcoal索引即可。
SQL> alter index IND_NAME rebuild partition p3;

Index altered.

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE

IND_NAME P4 20 USABLE

2.分区含有数据，加update indexes
SQL> alter table part_test move partition p3 update indexes;

Table altered.

加上update indexes后，两种索引都正常。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE

IND_NAME P3 15 USABLE
IND_NAME P4 20 USABLE

小结：分区中含有数据，move分区会造成global和分区对应local索引失效；加上update indexes参数能够避免这个问题。

五、truncate partition
1.分区含有数据，不加update indexes
SQL> alter table part_test truncate partition p3;

Table truncated.

global索引失效不可用
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD UNUSABLE

local索引依然可用
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS

------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE
IND_NAME P4 20 USABLE

2.分区含有数据，加update indexes
SQL> insert into part_test values(11,'lucy'); --落在p3分区

1 row created.

SQL> commit;

Commit complete.

重建global索引
SQL> alter index IND_ID rebuild;

Index altered.

SQL> alter table part_test truncate partition p3 update indexes;

Table truncated.
加上update indexes后，global索引不再失效。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE

IND_NAME P4 20 USABLE

小结：分区表中含有数据，truncate partition会造成global索引失效;truncate partition加上update indexes可用避免这种情况。

六、 SPLIT PARTITION （一个分区分裂为多个分区 )
准备实验分区

SQL> alter table part_test add partition max_part values less than(maxvalue);

Table altered.

SQL> insert into part_test values(21,'john');

1 row created.

SQL> insert into part_test values(30,'dog');

1 row created.

SQL> commit;

Commit complete.

SQL> select * from part_test partition(MAX_PART);

ID NAME
---------- ----------
21 john
30 dog

1.不加update indexes
SQL> alter table part_test split partition max_part at (25) into (partition p5,partition max_part );

Table altered.

global索引不可用
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD UNUSABLE

local索引，原分区和分裂出的新分区都不可用。
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS

------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE
IND_NAME P4 20 USABLE
IND_NAME P5 25 UNUSABLE

IND_NAME MAX_PART MAXVALUE UNUSABLE

2.加update indexes

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE

IND_NAME P4 20 USABLE
IND_NAME MAX_PART MAXVALUE USABLE

SQL> alter table part_test split partition max_part at (25) into (partition p5,partition max_part ) update indexes;

Table altered.

加update indexes后，分裂不会造成索引失效。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS

------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE
IND_NAME P4 20 USABLE
IND_NAME P5 25 USABLE

IND_NAME MAX_PART MAXVALUE USABLE

小结：split partition操作，会使global索引和local索引的原分区和分裂出的新分区都不可用。加上update indexes解决这个问题。

七、 merge partition( 合并分区 )
现有分区
SQL> select table_name,partition_name,HIGH_VALUE from dba_tab_partitions where table_name='PART_TEST' order by PARTITION_NAME;

TABLE_NAME PARTITION_NAME HIGH_VALUE
------------------------------ ------------------------------ --------------------------------------------------------------------------------
PART_TEST MAX_PART MAXVALUE
PART_TEST P1 5
PART_TEST P3 15
PART_TEST P4 20
PART_TEST P5 25

p4和p5分区都含有数据，把他们合并为p6分区

1.不加update indexes
SQL> alter table PART_TEST merge partitions p4,p5 into partition p6;

Table altered.

global索引失效
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD UNUSABLE

新生成的分区的索引是失效的。
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE

IND_NAME P6 25 UNUSABLE
IND_NAME MAX_PART MAXVALUE USABLE

2.加update indexes
SQL> alter table PART_TEST merge partitions p4,p5 into partition p6 update indexes;

Table altered.

global和local索引都正常。
SQL> SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS

------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE
IND_NAME P6 25 USABLE
IND_NAME MAX_PART MAXVALUE USABLE

八、COALESCE PARTITION （收缩表分区）
仅能应用在hash 分区表，执行一次就少一个分区，如一个hash分区表含有3个分区，COALESCE PARTITION 一次就成了2个分区。
create table part_hash(id number,name nvarchar2(20))
partition by hash(id)
(
partition p1,
partition p2,
partition p3,
partition p4
);

create index ind_hash_id on part_hash(id);
create index ind_hash_name on part_hash(name) local;

insert into part_hash values(1,'tom');
insert into part_hash values(2,'lucy');
insert into part_hash values(3,'john');
insert into part_hash values(4,'kate');

索引信息
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_HASH_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_HASH_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_HASH_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_HASH_NAME P1 USABLE

IND_HASH_NAME P2 USABLE
IND_HASH_NAME P3 USABLE
IND_HASH_NAME P4 USABLE

表信息
SQL> select table_name,partition_name,HIGH_VALUE from dba_tab_partitions where table_name='PART_HASH' order by table_name,partition_name;

TABLE_NAME PARTITION_NAME HIGH_VALUE
------------------------------ ------------------------------ --------------------------------------------------------------------------------
PART_HASH P1
PART_HASH P2
PART_HASH P3
PART_HASH P4

1.下面进行coalesce partiton操作,不加update index
SQL> alter table PART_HASH coalesce partition;

Table altered.

操作引起global和local索引失效。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_HASH_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_HASH_ID DOWNLOAD UNUSABLE

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_HASH_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_HASH_NAME P1 USABLE

IND_HASH_NAME P2 UNUSABLE
IND_HASH_NAME P3 USABLE

修复索引
alter index IND_HASH_ID rebuild;
alter index IND_HASH_NAME rebuild partition p2;

2.加update indexes
SQL> alter table PART_HASH coalesce partition update indexes;

Table altered.

操作后，索引没有失效。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_HASH_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_HASH_ID DOWNLOAD VALID

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_HASH_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_HASH_NAME P1 USABLE
IND_HASH_NAME P2 USABLE

九、 exchange partition
exchange partition是表与表，表与分区，分区与分区数据交换的命令，不是将表转换成分区，或者分区转换成表

依然用上面的part_test分区表
SQL> select * from PART_TEST partition(p6);

ID NAME
---------- ----------
18 ok
21 john
21 jodan

新建一个表exchange_test
create table exchange_test(id number,name varchar2(10));

下面把PART_TEST partition(p6)中的数据exchange到exchange_test中
SQL> alter table PART_TEST exchange partition p6 with table exchange_test;

Table altered.

p6分区没有数据了
SQL> select * from PART_TEST partition(p6);

no rows selected

数据跑到了exchange_test中。
SQL> select * from exchange_test;

ID NAME
---------- ----------
18 ok
21 john
21 jodan

看分区表索引情况，global索引失效，p6上的local索引失效。
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD UNUSABLE

SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE

IND_NAME P3 15 USABLE
IND_NAME P6 25 UNUSABLE
IND_NAME MAX_PART MAXVALUE USABLE

恢复索引
alter index IND_ID rebuild;
alter index IND_NAME rebuild partition p6;

2.加update indexes
SQL> alter table PART_TEST exchange partition p6 with table exchange_test update indexes;

数据又回去了
SQL> select * from exchange_test;

no rows selected

SQL> select * from PART_TEST partition(p6);

ID NAME
---------- ----------
18 ok

21 john
21 jodan

看索引情况
global索引正常
SQL> select INDEX_NAME,TABLE_OWNER,STATUS from dba_indexes where index_name='IND_ID';

INDEX_NAME TABLE_OWNER STATUS
------------------------------ ------------------------------ --------
IND_ID DOWNLOAD VALID

local索引还是失效了。
SQL> select INDEX_NAME,PARTITION_NAME,HIGH_VALUE,STATUS from dba_ind_partitions where INDEX_NAME='IND_NAME';

INDEX_NAME PARTITION_NAME HIGH_VALUE STATUS
------------------------------ ------------------------------ -------------------------------------------------------------------------------- --------
IND_NAME P1 5 USABLE
IND_NAME P3 15 USABLE

IND_NAME P6 25 UNUSABLE
IND_NAME MAX_PART MAXVALUE USABLE

小结：加上update indexes，global索引不失效，但是local 索引依然失效。

十、重命名分区和 local 索引名字都不会影响索引状态
SQL> alter table part_test rename partition P6 to p5;
SQL> alter index IND_NAME rename partition P6 to p5;

注意：
local 索引不能整个 rebuild ，要 rebuild 索引分区。
SQL> alter index IND_NAME rebuild;
alter index IND_NAME rebuild
*
ERROR at line 1:
ORA-14086: a partitioned index may not be rebuilt as a whole

SQL> alter index IND_NAME rebuild partition MAX_PART;

Index altered.

如果local索引大面积不可用，那么用动态sql生成重语句吧
SQL> select 'alter index INDEX_NAME rebuild partition '||PARTITION_NAME||';' from user_ind_partitions where INDEX_NAME='INDEX_NAME' and STATUS='UNUSABLE';