spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Takeshi Yamamuro (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SPARK-15247) sqlCtx.read.parquet yields at least n_executors * n_cores tasks
Date Thu, 12 May 2016 10:35:12 GMT

    [ https://issues.apache.org/jira/browse/SPARK-15247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15281408#comment-15281408
] 

Takeshi Yamamuro commented on SPARK-15247:
------------------------------------------

I found that `ParquetRelation` internally checks a schema by using `SparkContext#parallelize`
with `defaultParallelism` (that is, _executors * n_cores by default), so `sqlCtx.read.parquet`
generates many tasks. One of workarounds is to set a suitable number at `spark.default.parallelism`.
See: https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetRelation.scala#L800

How big does this issue make performance degradation?

> sqlCtx.read.parquet yields at least n_executors * n_cores tasks
> ---------------------------------------------------------------
>
>                 Key: SPARK-15247
>                 URL: https://issues.apache.org/jira/browse/SPARK-15247
>             Project: Spark
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 1.6.0
>            Reporter: Johnny W.
>
> sqlCtx.read.parquet always yields at least n_executors * n_cores tasks, even though this
is only 1 very small file
> This issue can increase the latency for small jobs.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message