[ https://issues.apache.org/jira/browse/SYSTEMML-1433?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15993416#comment-15993416 ]
Glenn Weidner commented on SYSTEMML-1433: ----------------------------------------- Fixed with [commit f380b52|https://github.com/apache/incubator-systemml/commit/f380b52a9d041798b268de8007140b418af1e0db]. > Unnecessarily blocking rdd and broadcast cleanup > ------------------------------------------------ > > Key: SYSTEMML-1433 > URL: https://issues.apache.org/jira/browse/SYSTEMML-1433 > Project: SystemML > Issue Type: Bug > Reporter: Matthias Boehm > Assignee: Matthias Boehm > Fix For: SystemML 0.14 > > > Our SparkExecutionContext has an internal configuration > {{ASYNCHRONOUS_VAR_DESTROY}} that indicates if rdds and broadcasts should be > cleaned up in an asynchronous manner (true by default). However, it is > incorrectly fed to the respective unpersist and destroy calls as a > {{blocking}} parameter, which essentially makes all cleanup calls unwillingly > synchronous. -- This message was sent by Atlassian JIRA (v6.3.15#6346)