Spark Job failing with exit status 15 - scala

Question

I am trying to run simple word count job in spark but I am getting exception while running job.
For more detailed output, check application tracking page:http://quickstart.cloudera:8088/proxy/application_1446699275562_0006/Then, click on links to logs of each attempt.
Diagnostics: Exception from container-launch.
Container id: container_1446699275562_0006_02_000001
Exit code: 15
Stack trace: ExitCodeException exitCode=15:
at org.apache.hadoop.util.Shell.runCommand(Shell.java:538)
at org.apache.hadoop.util.Shell.run(Shell.java:455)
at org.apache.hadoop.util.Shell$ShellCommandExecutor.execute(Shell.java:715)
at org.apache.hadoop.yarn.server.nodemanager.DefaultContainerExecutor.launchContainer(DefaultContainerExecutor.java:211)
at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:302)
at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:82)
at java.util.concurrent.FutureTask.run(FutureTask.java:262)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:745)
Container exited with a non-zero exit code 15
Failing this attempt. Failing the application.
ApplicationMaster host: N/A
ApplicationMaster RPC port: -1
queue: root.cloudera
start time: 1446910483956
final status: FAILED
tracking URL: http://quickstart.cloudera:8088/cluster/app/application_1446699275562_0006
user: cloudera
Exception in thread "main" org.apache.spark.SparkException: Application finished with failed status
at org.apache.spark.deploy.yarn.Client.run(Client.scala:626)
at org.apache.spark.deploy.yarn.Client$.main(Client.scala:651)
at org.apache.spark.deploy.yarn.Client.main(Client.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:606)
at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:569)
at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:166)
at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:189)
at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:110)
at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
I checked log from following command
yarn logs -applicationId application_1446699275562_0006
Here is log
15/11/07 07:35:09 ERROR yarn.ApplicationMaster: User class threw exception: Output directory hdfs://quickstart.cloudera:8020/user/cloudera/WordCountOutput already exists
org.apache.hadoop.mapred.FileAlreadyExistsException: Output directory hdfs://quickstart.cloudera:8020/user/cloudera/WordCountOutput already exists
at org.apache.hadoop.mapred.FileOutputFormat.checkOutputSpecs(FileOutputFormat.java:132)
at org.apache.spark.rdd.PairRDDFunctions.saveAsHadoopDataset(PairRDDFunctions.scala:1053)
at org.apache.spark.rdd.PairRDDFunctions.saveAsHadoopFile(PairRDDFunctions.scala:954)
at org.apache.spark.rdd.PairRDDFunctions.saveAsHadoopFile(PairRDDFunctions.scala:863)
at org.apache.spark.rdd.RDD.saveAsTextFile(RDD.scala:1290)
at org.com.td.sparkdemo.spark.WordCount$.main(WordCount.scala:23)
at org.com.td.sparkdemo.spark.WordCount.main(WordCount.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:606)
at org.apache.spark.deploy.yarn.ApplicationMaster$$anon$2.run(ApplicationMaster.scala:480)
15/11/07 07:35:09 INFO yarn.ApplicationMaster: Final app status: FAILED, exitCode: 15, (reason: User class threw exception: Output directory hdfs://quickstart.cloudera:8020/user/cloudera/WordCountOutput already exists)
15/11/07 07:35:14 ERROR yarn.ApplicationMaster: SparkContext did not initialize after waiting for 100000 ms. Please check earlier log output for errors. Failing the application.
Exception clearly indicates that WordCountOutput directory already exists but I made sure that directory is not there before running job.
Why I am getting this error even though directory was not there before running my job?

Answer 1

I came across same issue and fixed it by adding below highlighted part.
SparkConf sparkConf = new SparkConf().setAppName("Sentiment Scoring").set("spark.hadoop.validateOutputSpecs", "true");
Thanks,
Sathish.

Answer 2

For us this was caused by "running beyond physical memory limits". After increasing executer memory, the issue was fixed.

Answer 3

This Error mostly occurs while you submitting the wrong spark parameter in the spark-submit command. Please check the configuration parameters. In my case, I failed to submit masternamenode address where I need to read resources from HDFS.
Before : Where yarn threw Error- 15
dfs://masternamenode/TESTCGNATDATA/
After: Able to run the application
hdfs://masternamenode/TESTCGNATDATA/