[英]how can I keep partition'number not change when I use window.partitionBy() function with spark/scala?
I have a RDD
, the RDD' partition
of result changes to 200 when I use window
,can I not change partition
when I use window
? 我有一个
RDD
中, RDD' partition
的结果变为200,当我使用window
,我不能改变partition
,当我使用window
?
This is my code: 这是我的代码:
val rdd= sc.parallelize(List(1,3,2,4,5,6,7,8),4)
val result = rdd.toDF("values").withColumn("csum", sum(col("values")).over(Window.partitionBy(col("values")))).rdd
println(result.getNumPartitions + "rdd2")
My input partition is 4, why result partition is 200? 我的输入分区是4,为什么结果分区是200?
I want my result partition to be also 4. 我希望我的结果分区也是4。
Is there any cleaner solution? 有没有更清洁的解决方案?
Note: As mentioned by @eliasah - it's not possible to avoid repartition when using window functions with spark
注意:如@eliasah所述 - 当使用带有spark的窗口函数时,无法避免重新分区
- Why result partition is 200?
为什么结果分区是200?
Spark doc The default value of spark.sql.shuffle.partitions
which Configures the number of partitions to use when shuffling data for joins or aggregations - is 200 Spark doc
spark.sql.shuffle.partitions
的默认值,用于配置在为连接或聚合重排数据时使用的分区数 - 是200
- How can I repartition to 4?
我怎样才能重新分配到4?
You can use: 您可以使用:
coalesce(4)
or 要么
repartition(4)
coalesce(numPartitions) Decrease the number of partitions in the RDD to numPartitions. coalesce(numPartitions)将RDD中的分区数减少为numPartitions。 Useful for running operations more efficiently after filtering down a large dataset.
过滤大型数据集后,可以更有效地运行操作。
repartition(numPartitions) Reshuffle the data in the RDD randomly to create either more or fewer partitions and balance it across them. repartition(numPartitions)随机重新调整RDD中的数据以创建更多或更少的分区并在它们之间进行平衡。 This always shuffles all data over the network.
这总是随机播放网络上的所有数据。
(also added this answer to https://stackoverflow.com/a/44384638/3415409 ) (还将此答案添加到https://stackoverflow.com/a/44384638/3415409 )
I was just reading about controlling the number of partitions when using groupBy aggregation, from https://jaceklaskowski.gitbooks.io/mastering-spark-sql/spark-sql-performance-tuning-groupBy-aggregation.html , it seems the same trick works with Window, in my code I'm defining a window like 我只是在阅读有关使用groupBy聚合时控制分区数量的文章,来自https://jaceklaskowski.gitbooks.io/mastering-spark-sql/spark-sql-performance-tuning-groupBy-aggregation.html ,看起来是一样的诀窍适用于Window,在我的代码中我定义了一个窗口
windowSpec = Window \
.partitionBy('colA', 'colB') \
.orderBy('timeCol') \
.rowsBetween(1, 1)
and then doing 然后做
next_event = F.lead('timeCol', 1).over(windowSpec)
and creating a dataframe via 并通过创建数据帧
df2 = df.withColumn('next_event', next_event)
and indeed, it has 200 partitions. 事实上,它有200个分区。 But, if I do
但是,如果我这样做
df2 = df.repartition(10, 'colA', 'colB').withColumn('next_event', next_event)
it has 10! 它有10个!
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.