[SPARK-21146] Master/Worker should handle and shutdown when any thread gets UncaughtException

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.1.1
• Fix Version/s: 2.3.0
• Component/s: Spark Core
• Assignee: Devaraj Kavali
• Reporter: Devaraj Kavali
• Created: 20/Jun/17 00:36
• Updated: 12/Jul/17 07:15
• Resolved: 12/Jul/17 07:15

Description

17/06/19 11:41:23 INFO Worker: Asked to launch executor app-20170619114055-0005/228 for ScalaSort
Exception in thread "dispatcher-event-loop-79" java.lang.OutOfMemoryError: unable to create new native thread
	at java.lang.Thread.start0(Native Method)
	at java.lang.Thread.start(Thread.java:714)
	at java.util.concurrent.ThreadPoolExecutor.addWorker(ThreadPoolExecutor.java:950)
	at java.util.concurrent.ThreadPoolExecutor.processWorkerExit(ThreadPoolExecutor.java:1018)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1160)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

I see in the logs that Worker's dispatcher-event got the above exception and the Worker keeps running without performing any functionality. And also Worker state changed from ALIVE to DEAD in Master's web UI.

worker-20170619150349-192.168.1.120-56175 	192.168.1.120:56175 	DEAD 	88 (41 Used) 	251.2 GB (246.0 GB Used)

I think Worker should handle and shutdown when any thread gets UncaughtException.

Issue Links

• duplicates: SPARK-21148 Set SparkUncaughtExceptionHandler to the Master
• links to: [Github] Pull Request #18357 (devaraj-kavali)