[ https://issues.apache.org/jira/browse/SPARK-4087?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14183676#comment-14183676 ]
Apache Spark commented on SPARK-4087: ------------------------------------- User 'davies' has created a pull request for this issue: https://github.com/apache/spark/pull/2933 > Only use broadcast for large tasks > ---------------------------------- > > Key: SPARK-4087 > URL: https://issues.apache.org/jira/browse/SPARK-4087 > Project: Spark > Issue Type: Improvement > Reporter: Davies Liu > Priority: Critical > > After we broadcast every tasks, some regressions are introduced because of > broadcast is not stable enough. > So we would like to only use broadcast for large tasks, which will keep the > same behaviour as 1.0 for most of the cases. -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org