CREATE SLICEGROUP

Description

Creates a group of distribution rules for the usage of range bucket tables and persist the rules to the gs_slicegroup system catalog.

Precautions

• Multiple slices cannot be distributed to the same bucket. When creating a slice group, you need to set bucketcnt to a value greater than or equal to the number of slices.
• When a slice group is created, a slice can only be distributed to only one DN.
• The number of bucketcnt allocated to a slice group cannot exceed 1024. A slice group is a library-level concept. The total number of buckets allocated to all slice groups in a database cannot exceed 4096.
• When creating a slice group, you cannot specify the distribution key type or the character set.
• The slice definition supports only the LESS THAN syntax but does not support START END and REFERENCE.
• If gs_dump is used to export data from a range bucket table and gs_restore is used to import data to the table, you need to manually create a slice group before import. When the audit function is enabled, you can search for the creation statement in the audit log.

Syntax

1
2
3

CREATE SLICEGROUP slicegroup_name 
    DISTRIBUTE BY RANGE ( column_type [, ...] )
    ( slice_less_than_item [, ...] ) BUCKETCNT bucket_cnt;

Parameters

• slicegroup_name

  Name of the distribution rule to be created. The name must be unique in the same database and different from the node group name.

  Value range: a string, which can contain lowercase letters (a-z), uppercase letters (A-Z), underscores (_), digits (0-9), and dollar signs ($). The value must start with a letter or underscore.

• column_type

  Distribution key type. A maximum of four distribution keys are supported.

• slice_less_than_item

  Slice definition using the LESS THAN syntax.

  For slice_less_than_item:
  SLICE name VALUES LESS THAN ({ expression | MAXVALUE } [, ...]) [ DATANODE datanode_name ]

  Value range of name: a string, which can contain lowercase letters (a-z), uppercase letters (A-Z), underscores (_), digits (0-9), and dollar signs ($), and must start with a letter or underscore.

• bucket_cnt

  Total number of buckets to be maintained by the distribution rule created.

  Value range: [1,1024]

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24

-- Create a slice group.
gaussdb=# CREATE SLICEGROUP sg DISTRIBUTE BY RANGE(INT,CHAR(5),TIMESTAMP,TEXT)
(SLICE s1 VALUES LESS THAN (1000,'3700x','2012-5-5 10:52:12','customer') DATANODE datanode3,
 SLICE s2 VALUES LESS THAN (1000,'3701x','2012-5-5 10:52:12','customer'),
 SLICE s3 VALUES LESS THAN (1000,'3702x','2012-5-5 10:52:12','customer')) bucketcnt 3;

-- Query the slice group.
gaussdb=# SELECT * FROM gs_slicegroup;
 slicename | type | strategy | bucketgroup | bucketcnt | sgroupoid | sindex | nodeid | redis_nodeid | bucket | slicecnt | contained | specified | sliceorder |                boundaries
     | bucketversion | bucketcsn | bucketxid | distributetype
-----------+------+----------+-------------+-----------+-----------+--------+--------+--------------+--------+----------+-----------+-----------+------------+--------------------------------------
-----+---------------+-----------+-----------+-----------------
 sg        | t    | r        |           1 |         3 |     16394 |      0 |     -1 |           -1 |     -1 |        3 | f         | f         |         -1 |
     |               |           |           | 23 1042 1114 25
 s1        | s    | r        |           1 |         1 |     16394 |      0 |      2 |           -1 |   2048 |       -1 | f         | t         |          0 | {1000,3700x,"2012-5-5 10:52:12",custo
mer} |               |           |           |
 s2        | s    | r        |           1 |         1 |     16394 |      0 |      0 |           -1 |   2049 |       -1 | f         | f         |          1 | {1000,3701x,"2012-5-5 10:52:12",custo
mer} |               |           |           |
 s3        | s    | r        |           1 |         1 |     16394 |      0 |      1 |           -1 |   2050 |       -1 | f         | f         |          2 | {1000,3702x,"2012-5-5 10:52:12",custo
mer} |               |           |           |
(4 rows)

-- Drop a slice group.
gaussdb=# DROP SLICEGROUP sg;

Helpful Links

DROP SLICEGROUP