求助：spark on yarn中client模式报错

如题，但是用cluster模式就可以正确运行。
环境是java1.8        hadoop2.7.3          spark1.7.3
附上yarn的日志
ubuntu@Master:/usr/soft/hadoop-2.7.3/logs$ spark-submit --class org.apache.spark.examples.SparkPi --master yarn --deploy-mode client --driver-memory 1G --executor-memory 1G --executor-cores 1 /usr/soft/spark-1.6.3/examples/target/spark-examples_2.10-1.6.3.jar 10
17/02/27 16:11:37 INFO util.Utils: Successfully started service 'sparkDriverActorSystem' on port 46860.
17/02/27 16:11:37 INFO spark.SparkEnv: Registering MapOutputTracker
17/02/27 16:11:37 INFO spark.SparkEnv: Registering BlockManagerMaster
17/02/27 16:11:37 INFO storage.DiskBlockManager: Created local directory at /tmp/blockmgr-6b35587e-ebb4-4468-a7ba-e8b734982a99
17/02/27 16:11:37 INFO storage.MemoryStore: MemoryStore started with capacity 511.1 MB
17/02/27 16:11:37 INFO spark.SparkEnv: Registering OutputCommitCoordinator
17/02/27 16:11:37 INFO server.AbstractConnector: Started [email protected]:4040
17/02/27 16:11:37 INFO util.Utils: Successfully started service 'SparkUI' on port 4040.
17/02/27 16:11:37 INFO ui.SparkUI: Started SparkUI at http://172.16.239.1:4040
17/02/27 16:11:37 INFO spark.HttpFileServer: HTTP File server directory is /tmp/spark-cbd08d7c-fc96-4d16-9cbb-257e2edb3728/httpd-4f06c896-7401-420a-8eb1-6e0be82e7dd6
17/02/27 16:11:38 INFO spark.HttpServer: Starting HTTP Server
17/02/27 16:11:38 INFO server.Server: jetty-8.y.z-SNAPSHOT
17/02/27 16:11:38 INFO server.AbstractConnector: Started [email protected]:45684
17/02/27 16:11:38 INFO util.Utils: Successfully started service 'HTTP file server' on port 45684.
17/02/27 16:11:38 INFO spark.SparkContext: Added JAR file:/usr/soft/spark-1.6.3/examples/target/spark-examples_2.10-1.6.3.jar at http://172.16.239.1:45684/jars/spark-examples_2.10-1.6.3.jar with timestamp 1488183098062
17/02/27 16:11:38 INFO client.RMProxy: Connecting to ResourceManager at Master/172.16.239.1:8032
17/02/27 16:11:38 INFO yarn.Client: Requesting a new application from cluster with 3 NodeManagers
17/02/27 16:11:38 INFO yarn.Client: Verifying our application has not requested more than the maximum memory capability of the cluster (8192 MB per container)
17/02/27 16:11:38 INFO yarn.Client: Will allocate AM container, with 896 MB memory including 384 MB overhead
17/02/27 16:11:38 INFO yarn.Client: Setting up container launch context for our AM
17/02/27 16:11:38 INFO yarn.Client: Setting up the launch environment for our AM container
17/02/27 16:11:38 INFO yarn.Client: Preparing resources for our AM container
17/02/27 16:11:40 INFO yarn.Client: Uploading resource file:/usr/soft/spark-1.6.3/assembly/target/scala-2.10/spark-assembly-1.6.3-hadoop2.7.3.jar -> hdfs://Master:9000/user/ubuntu/.sparkStaging/application_1488181713860_0003/spark-assembly-1.6.3-hadoop2.7.3.jar
17/02/27 16:11:47 INFO yarn.Client: Uploading resource file:/tmp/spark-cbd08d7c-fc96-4d16-9cbb-257e2edb3728/__spark_conf__7095116935778215646.zip -> hdfs://Master:9000/user/ubuntu/.sparkStaging/application_1488181713860_0003/__spark_conf__7095116935778215646.zip
17/02/27 16:11:47 INFO spark.SecurityManager: Changing view acls to: ubuntu
17/02/27 16:11:47 INFO spark.SecurityManager: Changing modify acls to: ubuntu
17/02/27 16:11:47 INFO spark.SecurityManager: SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(ubuntu); users with modify permissions: Set(ubuntu)
17/02/27 16:11:47 INFO yarn.Client: Submitting application 3 to ResourceManager
17/02/27 16:11:47 INFO impl.YarnClientImpl: Submitted application application_1488181713860_0003
17/02/27 16:11:48 INFO yarn.Client: Application report for application_1488181713860_0003 (state: ACCEPTED)
17/02/27 16:11:48 INFO yarn.Client: 
 client token: N/A
 diagnostics: N/A
 ApplicationMaster host: N/A
 ApplicationMaster RPC port: -1
 queue: default
 start time: 1488183107661
 final status: UNDEFINED
 tracking URL: http://Master:8088/proxy/application_1488181713860_0003/
 user: ubuntu
17/02/27 16:12:05 INFO yarn.Client: Application report for application_1488181713860_0003 (state: ACCEPTED)
17/02/27 16:12:06 INFO yarn.Client: Application report for application_1488181713860_0003 (state: ACCEPTED)
17/02/27 16:12:07 INFO yarn.Client: Application report for application_1488181713860_0003 (state: FAILED)
17/02/27 16:12:07 INFO yarn.Client: 
 client token: N/A
 diagnostics: Application application_1488181713860_0003 failed 2 times due to AM Container for appattempt_1488181713860_0003_000002 exited with  exitCode: -103
Container killed on request. Exit code is 143
Container exited with a non-zero exit code 143
Failing this attempt. Failing the application.
 ApplicationMaster host: N/A
 ApplicationMaster RPC port: -1
 queue: default
 start time: 1488183107661
 final status: FAILED
 tracking URL: http://Master:8088/cluster/app/application_1488181713860_0003
 user: ubuntu
17/02/27 16:12:07 INFO yarn.Client: Deleting staging directory .sparkStaging/application_1488181713860_0003
17/02/27 16:12:07 ERROR spark.SparkContext: Error initializing SparkContext.
org.apache.spark.SparkException: Yarn application has already ended! It might have been killed or unable to launch application master.
at org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.waitForApplication(YarnClientSchedulerBackend.scala:124)
at org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.start(YarnClientSchedulerBackend.scala:64)
at org.apache.spark.scheduler.TaskSchedulerImpl.start(TaskSchedulerImpl.scala:144)
at org.apache.spark.SparkContext.<init>(SparkContext.scala:530)
at org.apache.spark.examples.SparkPi$.main(SparkPi.scala:29)
at org.apache.spark.examples.SparkPi.main(SparkPi.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:731)
at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:181)
at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:206)
at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:121)
at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
17/02/27 16:12:07 INFO handler.ContextHandler: stopped o.s.j.s.ServletContextHandler{/stages/stage/kill,null}17/02/27 16:12:07 INFO handler.ContextHandler: stopped o.s.j.s.ServletContextHandler{/jobs,null}
17/02/27 16:12:07 INFO ui.SparkUI: Stopped Spark web UI at http://172.16.239.1:4040
17/02/27 16:12:07 INFO cluster.YarnClientSchedulerBackend: Shutting down all executors
17/02/27 16:12:07 INFO cluster.YarnClientSchedulerBackend: Asking each executor to shut down
17/02/27 16:12:07 INFO cluster.YarnClientSchedulerBackend: Stopped
17/02/27 16:12:07 INFO spark.MapOutputTrackerMasterEndpoint: MapOutputTrackerMasterEndpoint stopped!
17/02/27 16:12:07 INFO storage.MemoryStore: MemoryStore cleared
17/02/27 16:12:07 INFO storage.BlockManager: BlockManager stopped
17/02/27 16:12:07 INFO storage.BlockManagerMaster: BlockManagerMaster stopped
17/02/27 16:12:07 WARN metrics.MetricsSystem: Stopping a MetricsSystem that is not running
17/02/27 16:12:08 INFO scheduler.OutputCommitCoordinator$OutputCommitCoordinatorEndpoint: OutputCommitCoordinator stopped!
17/02/27 16:12:08 INFO spark.SparkContext: Successfully stopped SparkContext
Exception in thread "main" org.apache.spark.SparkException: Yarn application has already ended! It might have been killed or unable to launch application master.
at org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.waitForApplication(YarnClientSchedulerBackend.scala:124)
at org.apache.spark.scheduler.cluster.YarnClientSchedulerBackend.start(YarnClientSchedulerBackend.scala:64)
at org.apache.spark.scheduler.TaskSchedulerImpl.start(TaskSchedulerImpl.scala:144)
at org.apache.spark.SparkContext.<init>(SparkContext.scala:530)
at org.apache.spark.examples.SparkPi$.main(SparkPi.scala:29)
at org.apache.spark.examples.SparkPi.main(SparkPi.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:498)
at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:731)
at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:181)
at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:206)
at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:121)
at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
17/02/27 16:12:08 INFO remote.RemoteActorRefProvider$RemotingTerminator: Shutting down remote daemon.
17/02/27 16:12:08 INFO remote.RemoteActorRefProvider$RemotingTerminator: Remote daemon shut down; proceeding with flushing remote transports.
17/02/27 16:12:08 INFO util.ShutdownHookManager: Shutdown hook called
17/02/27 16:12:08 INFO util.ShutdownHookManager: Deleting directory /tmp/spark-cbd08d7c-fc96-4d16-9cbb-257e2edb3728/httpd-4f06c896-7401-420a-8eb1-6e0be82e7dd6
17/02/27 16:12:08 INFO util.ShutdownHookManager: Deleting directory /tmp/spark-cbd08d7c-fc96-4d16-9cbb-257e2edb3728
节点日志
17/02/27 00:28:49 INFO yarn.ApplicationMaster: Registered signal handlers for [TERM, HUP, INT]
17/02/27 00:28:51 INFO yarn.ApplicationMaster: ApplicationAttemptId: appattempt_1488183797412_0002_000001
17/02/27 00:28:52 INFO yarn.ApplicationMaster: Final app status: FAILED, exitCode: 16
17/02/27 00:28:52 ERROR yarn.ApplicationMaster: RECEIVED SIGNAL 15: SIGTERM
17/02/27 00:28:52 INFO util.ShutdownHookManager: Shutdown hook called