[
https://issues.apache.org/jira/browse/SPARK-18297?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15643795#comment-15643795
]
liuhongqiang edited comment on SPARK-18297 at 11/7/16 10:41 AM:
----------------------------------------------------------------
I use a shared singleton SparkContext in product environment.
The issue is SparkContext can't finish crrectly in asynchronous thread.
was (Author: smallyard):
I use a shared singleton SparkContext in product environment.
> Fail if SparkContext run a new Thread in yarn-cluster
> -----------------------------------------------------
>
> Key: SPARK-18297
> URL: https://issues.apache.org/jira/browse/SPARK-18297
> Project: Spark
> Issue Type: Bug
> Components: YARN
> Reporter: liuhongqiang
>
> program:
> public static void main(String[] args) {
> Executors.newSingleThreadScheduledExecutor().scheduleAtFixedRate(new
> Thread(new Runnable() {
> @Override
> public void run() {
> SparkConf conf = new SparkConf();
> conf.setAppName("SparkDemo");
> JavaSparkContext sparkContext = new JavaSparkContext(conf);
> JavaRDD<String> array =
> sparkContext.parallelize(Lists.newArrayList("1", "2", "3", "4"));
> System.out.println(array.count());
> }
> }), 0, 5000, TimeUnit.MILLISECONDS);
> }
> log:
> 16/11/02 11:16:47 INFO yarn.ApplicationMaster: Starting the user application
> in a separate Thread
> 16/11/02 11:16:47 INFO yarn.ApplicationMaster: Waiting for spark context
> initialization
> 16/11/02 11:16:47 INFO yarn.ApplicationMaster: Waiting for spark context
> initialization ...
> 16/11/02 11:16:47 INFO yarn.ApplicationMaster: Final app status: SUCCEEDED,
> exitCode: 0
> problem:
> mainMethod.invoke(null, userArgs.toArray)
> finish(FinalApplicationStatus.SUCCEEDED, ApplicationMaster.EXIT_SUCCESS)
> main method was finished, but sub thread may be not finished.
> so should not invoke finish to shutdown dirver thread.
--
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
