repartition

Returns a new DataFrame partitioned by the given partitioning expressions. The resulting DataFrame is hash partitioned.

Syntax

repartition(numPartitions: Union[int, "ColumnOrName"], *cols: "ColumnOrName")

Parameters

Parameter	Type	Description
numPartitions	int	can be an int to specify the target number of partitions or a Column. If it is a Column, it will be used as the first partitioning column. If not specified, the default number of partitions is used.
cols	str or Column	partitioning columns.

Returns

DataFrame: Repartitioned DataFrame.

Examples

from pyspark.sql import functions as sf
df = spark.range(0, 64, 1, 9).withColumn(
    "name", sf.concat(sf.lit("name_"), sf.col("id").cast("string"))
).withColumn(
    "age", sf.col("id") - 32
)
df.repartition(10).select(
    sf.spark_partition_id().alias("partition")
).distinct().sort("partition").show()
# +---------+
# |partition|
# +---------+
# |        0|
# ...
# |        9|
# +---------+

df.repartition(7, "age").select(
    sf.spark_partition_id().alias("partition")
).distinct().sort("partition").show()
# +---------+
# |partition|
# +---------+
# |        0|
# ...
# |        6|
# +---------+