[SPARK-8754] [YARN] YarnClientSchedulerBackend doesn't stop gracefull…

…y in failure conditions

In YarnClientSchedulerBackend.stop(), added a check for monitorThread.

Author: Devaraj K <devaraj@apache.org>

Closes #7153 from devaraj-kavali/master and squashes the following commits:

66be9ad [Devaraj K] https://issues.apache.org/jira/browse/SPARK-8754 YarnClientSchedulerBackend doesn't stop gracefully in failure conditions

(cherry picked from commit 792fcd8)
Signed-off-by: Andrew Or <andrew@databricks.com>

yarn/src/main/scala/org/apache/spark/scheduler/cluster/YarnClientSchedulerBackend.scala

@@ -147,7 +147,9 @@ private[spark] class YarnClientSchedulerBackend(
    */
   override def stop() {
     assert(client != null, "Attempted to stop this scheduler before starting it!")
-    monitorThread.interrupt()
+    if (monitorThread != null) {
+      monitorThread.interrupt()
+    }
     super.stop()
     client.stop()
     logInfo("Stopped")