Mailing-List: contact issues-help@spark.apache.org; run by ezmlm
Precedence: bulk
Date: Sun, 23 Jul 2017 15:25:00 +0000 (UTC)
From: "Wenchen Fan (JIRA)" <jira@apache.org>
To: issues@spark.apache.org
Message-ID: <JIRA.13075309.1495846781000.311054.1500823500528@Atlassian.JIRA>
In-Reply-To: <JIRA.13075309.1495846781000@Atlassian.JIRA>
References: <JIRA.13075309.1495846781000@Atlassian.JIRA> <JIRA.13075309.1495846781097@jira-lw-us.apache.org>
Subject: [jira] [Resolved] (SPARK-20904) Task failures during shutdown cause
 problems with preempted executors
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Sun, 23 Jul 2017 15:25:07 -0000


     [ https://issues.apache.org/jira/browse/SPARK-20904?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Wenchen Fan resolved SPARK-20904.
---------------------------------
       Resolution: Fixed
    Fix Version/s: 2.3.0
                   2.2.1

Issue resolved by pull request 18594
[https://github.com/apache/spark/pull/18594]

> Task failures during shutdown cause problems with preempted executors
> ---------------------------------------------------------------------
>
>                 Key: SPARK-20904
>                 URL: https://issues.apache.org/jira/browse/SPARK-20904
>             Project: Spark
>          Issue Type: Bug
>          Components: Spark Core, YARN
>    Affects Versions: 1.6.0
>            Reporter: Marcelo Vanzin
>             Fix For: 2.2.1, 2.3.0
>
>
> Spark runs tasks in a thread pool that uses daemon threads in each executor. That means that when the JVM gets a signal to shut down, those tasks keep running.
> Now when YARN preempts an executor, it sends a SIGTERM to the process, triggering the JVM shutdown. That causes shutdown hooks to run which may cause user code running in those tasks to fail, and report task failures to the driver. Those failures are then counted towards the maximum number of allowed failures, even though in this case we don't want that because the executor was preempted.
> So we need a better way to handle that situation.


--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

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