Add YarnClientClusterScheduler and Backend.

With this scheduler, the user application is launched locally,
While the executor will be launched by YARN on remote nodes.

This enables spark-shell to run upon YARN.

1 files changed, 25 insertions, 0 deletions

diff --git a/core/src/main/scala/org/apache/spark/SparkContext.scala b/core/src/main/scala/org/apache/spark/SparkContext.scala
index 42b2985b50..3a80241daa 100644
--- a/core/src/main/scala/org/apache/spark/SparkContext.scala
+++ b/core/src/main/scala/org/apache/spark/SparkContext.scala
@@ -226,6 +226,31 @@ class SparkContext(
         scheduler.initialize(backend)
         scheduler
 
+      case "yarn-client" =>
+        val scheduler = try {
+          val clazz = Class.forName("org.apache.spark.scheduler.cluster.YarnClientClusterScheduler")
+          val cons = clazz.getConstructor(classOf[SparkContext])
+          cons.newInstance(this).asInstanceOf[ClusterScheduler]
+
+        } catch {
+          case th: Throwable => {
+            throw new SparkException("YARN mode not available ?", th)
+          }
+        }
+
+        val backend = try {
+          val clazz = Class.forName("org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend")
+          val cons = clazz.getConstructor(classOf[ClusterScheduler], classOf[SparkContext])
+          cons.newInstance(scheduler, this).asInstanceOf[CoarseGrainedSchedulerBackend]
+        } catch {
+          case th: Throwable => {
+            throw new SparkException("YARN mode not available ?", th)
+          }
+        }
+
+        scheduler.initialize(backend)
+        scheduler
+
       case MESOS_REGEX(mesosUrl) =>
         MesosNativeLibrary.load()
         val scheduler = new ClusterScheduler(this)