Help Center> Data Lake Insight> Spark SQL Syntax Reference> Tables> Partition-related Syntax> Deleting a Partition
Updated on 2023-11-03 GMT+08:00
View PDF

Deleting a Partition

Function

Deletes one or more partitions from a partitioned table.

Precautions

  • The table in which partitions are to be deleted must exist. Otherwise, an error is reported.
  • The to-be-deleted partition must exist. Otherwise, an error is reported. To avoid this error, add IF EXISTS in this statement.

Syntax

1
2
3
 
ALTER TABLE [db_name.]table_name
  DROP [IF EXISTS]
  PARTITION partition_spec1[,PARTITION partition_spec2,...];

Keyword

  • DROP: deletes a partition.
  • IF EXISTS: The partition to be deleted must exist. Otherwise, an error is reported.
  • PARTITION: specifies the partition to be deleted

Parameters

Table 1 Parameter description

Parameter

Description

db_name

Database name that contains letters, digits, and underscores (_). It cannot contain only digits and cannot start with an underscore (_).

table_name

Table name of a database that contains letters, digits, and underscores (_). It cannot contain only digits and cannot start with an underscore (_). The matching rule is ^(?!_)(?![0-9]+$)[A-Za-z0-9_$]*$. If special characters are required, use single quotation marks ('') to enclose them.

partition_specs

Partition information, in the format of "key=value", where key indicates the partition field and value indicates the partition value. In a table partitioned using multiple fields, if you specify all the fields of a partition name, only the partition is deleted; if you specify only some fields of a partition name, all matching partitions will be deleted. By default, the partition_specs parameter contains (). For example: PARTITION (dt='2009-09-09',city='xxx')

Example

To delete the dt = '2008-08-08', city = 'xxx' partition in the student table, run the following statement:

1
2
3
 
ALTER TABLE student
  DROP
  PARTITION (dt = '2008-08-08', city = 'xxx');
Parent topic: Partition-related Syntax