[
https://issues.apache.org/jira/browse/SPARK-3478?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14129503#comment-14129503
]
Apache Spark commented on SPARK-3478:
-------------------------------------
User 'davies' has created a pull request for this issue:
https://github.com/apache/spark/pull/2351
> Profile Python tasks stage by stage in worker
> ---------------------------------------------
>
> Key: SPARK-3478
> URL: https://issues.apache.org/jira/browse/SPARK-3478
> Project: Spark
> Issue Type: New Feature
> Components: PySpark
> Reporter: Davies Liu
> Assignee: Davies Liu
>
> The Python code in driver is easy to profile by users, but the code run in
> worker is distributed in clusters, is not easy to profile by users.
> So we need a way to do the profiling in worker and aggregate all the result
> together for users.
> This also can be used to analys the bottleneck in PySpark.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
