Mailing-List: contact dev-help@hive.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@hive.apache.org
Date: Tue, 25 Nov 2014 09:37:12 +0000 (UTC)
From: "Chengxiang Li (JIRA)" <jira@apache.org>
To: hive-dev@hadoop.apache.org
Message-ID: <JIRA.12757643.1416904513000.14325.1416908232316@Atlassian.JIRA>
In-Reply-To: <JIRA.12757643.1416904513000@Atlassian.JIRA>
References: <JIRA.12757643.1416904513000@Atlassian.JIRA>
 <JIRA.12757643.1416904513500@arcas>
Subject: [jira] [Updated] (HIVE-8959) SparkSession is not closed until JVM
 exit.[Spark Branch]
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/HIVE-8959?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Chengxiang Li updated HIVE-8959:
--------------------------------
    Attachment: HIVE-8959.1-spark.patch

> SparkSession is not closed until JVM exit.[Spark Branch]
> --------------------------------------------------------
>
>                 Key: HIVE-8959
>                 URL: https://issues.apache.org/jira/browse/HIVE-8959
>             Project: Hive
>          Issue Type: Sub-task
>          Components: Spark
>            Reporter: Chengxiang Li
>            Assignee: Chengxiang Li
>              Labels: Spark-M3
>         Attachments: HIVE-8959.1-spark.patch
>
>
> During unit test, SparkSession is closed by Runtime shutdownHook, which means it's closed until JVM exist. During unit test suite, each qfile, as a single test case, would reset SessionState, which lead to a new Sparksession is created for each qfile. As we know that, RemoteSparkClient is SparkSession specified, so more and more executors is launched during unit test until blocked by no more resources.


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