“User did not initialize spark context” Error when using Scala code in SPARK YARN Cluster mode

Question

To narrow down the problem I removed other class dependency and I have this clean code:

object LoaderProcessor extends App {

val logger = LoggerFactory.getLogger(this.getClass())
execute()

def execute(): Unit = {

val spark = get_spark()
import spark.implicits._

var df = spark.read
  .format("csv")
  .option("delimiter", ",")
  .option("header", true)
  .option("inferSchema", "true")
  .option("timestampFormat", "yyyy/MM/dd HH:mm:ss")
  .load(args(2))

df = df.withColumn("zs_source", lit(1)) //the only operation on dataframe

val o_file = Config().getString("myapp.dataFolder") + "/8/1/data.csv"
logger.info("Writing output to: {}", o_file)

df.write.mode("overwrite")
.option("header", "true").csv(o_file)

}

def get_spark(): SparkSession = {
val env = System.getenv("MYAPP_ENV")
var spark:SparkSession = null
if (env == null || env == "dev_local") {
  spark = org.apache.spark.sql.SparkSession.builder
    .master("local")
    .appName("MyApp")
    .getOrCreate;
}else{
  spark = org.apache.spark.sql.SparkSession.builder
    .appName("MyApp")
    //.enableHiveSupport()
    .getOrCreate;
}
spark.sparkContext.setCheckpointDir(Config().getString("myapp.rddcp"))
return spark
}
}

It works well in client mode. Could not figure out the problem. I have my clusters on HDInsight.

Also noticed that the "write" operation keeps writing on output folder like this:

part-00000-3e9566ae-c13c-468a-8732-e7b8a8df5335-c000.csv

and then in few seconds:

part-00000-4f4979a0-d9f9-481b-aac4-115e63b9f59c-c000.csv

8/12/01 15:08:53 INFO ApplicationMaster: Starting the user application in a separate Thread 18/12/01 15:08:53 INFO ApplicationMaster: Waiting for spark context initialization... 18/12/01 15:08:55 INFO Config$: Environment: dev 18/12/01 15:08:55 ERROR ApplicationMaster: Uncaught exception: java.lang.IllegalStateException: User did not initialize spark context! at org.apache.spark.deploy.yarn.ApplicationMaster.runDriver(ApplicationMaster.scala:510) at org.apache.spark.deploy.yarn.ApplicationMaster.org$apache$spark$deploy$yarn$ApplicationMaster$$runImpl(ApplicationMaster.scala:345) at org.apache.spark.deploy.yarn.ApplicationMaster$$anonfun$run$2.apply$mcV$sp(ApplicationMaster.scala:260) at org.apache.spark.deploy.yarn.ApplicationMaster$$anonfun$run$2.apply(ApplicationMaster.scala:260) at org.apache.spark.deploy.yarn.ApplicationMaster$$anonfun$run$2.apply(ApplicationMaster.scala:260) at org.apache.spark.deploy.yarn.ApplicationMaster$$anon$5.run(ApplicationMaster.scala:815) at java.security.AccessController.doPrivileged(Native Method) at javax.security.auth.Subject.doAs(Subject.java:422) at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1869) at org.apache.spark.deploy.yarn.ApplicationMaster.doAsUser(ApplicationMaster.scala:814) at org.apache.spark.deploy.yarn.ApplicationMaster.run(ApplicationMaster.scala:259) at org.apache.spark.deploy.yarn.ApplicationMaster$.main(ApplicationMaster.scala:839) at org.apache.spark.deploy.yarn.ApplicationMaster.main(ApplicationMaster.scala)

spark-submit --master yarn --deploy-mode cluster --jars "wasb://xx@yy/zs/jars/config-1.3.1.jar" --class myapp.LoaderProcessor "wasb://xx@yy/zs/jars/myapp.jar" l 8 /data/8_data.csv 1 , true false -->PROBLEM

spark-submit --deploy-mode client --jars "wasb://xx@yy/zs/jars/config-1.3.1.jar" --class myapp.LoaderProcessor "wasb://xx@yy/zs/jars/myapp.jar" l 8 /data/8_data.csv 1 , true false -->WORKS!!!

Arnon Rotem-Gal-Oz Arnon Rotem-Gal-Oz · Accepted Answer · 2018-12-02T07:30:55

edit: updated per our exchange in the comments

the problem is that you're always creating a local context with the if (env == null || env == "dev_local") (MYAPP_ENV is null in distributed env)

“User did not initialize spark context” Error when using Scala code in SPARK YARN Cluster mode

1 Answers