操作分区表对global和local索引的影响 (三）
以下转自：http://blog.chinaunix.net/uid-23284114-id-3421922.html 作者：十字螺丝钉
操作分区表对global和local索引的影响 (二）请参考 http://bfc99.blog.51cto.com/265386/1436443
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
操作分区表对global和local索引的影响 (四）请参考 http://bfc99.blog.51cto.com/265386/1436446