How to check a correct install of spark? ( Whether spark is working as expected!)

[#1]

You may want to launch a spark-shell, check the version 'sc.version', check the instantiation of contexts/session and run some SQL queries.

[#2]

A good way to sanity check Spark is to start Spark shell with YARN (spark-shell --master yarn) and run something like this:

val x = sc.textFile("some hdfs path to a text file or directory of text files")

x.count()

This will basically do a distributed line count.

If that looks good, another sanity check is for Hive integration. Run spark-sql (spark-sql --master yarn) and try to query a table that you know can be queried via Hive.

[#3]

The Spark version will be displayed in the console log output...

[#4 · Accepted solution]

# Check Commands
# --------------

# Spark Scala
# -----------

# Optionally export Spark Home
export SPARK_HOME=/usr/hdp/current/spark-client

# Spark submit example in local mode
spark-submit --class org.apache.spark.examples.SparkPi --driver-memory 512m --executor-memory 512m --executor-cores 1 $SPARK_HOME/lib/spark-examples*.jar 10

# Spark submit example in client mode
spark-submit --class org.apache.spark.examples.SparkPi --master yarn-client --num-executors 3 --driver-memory 512m --executor-memory 512m --executor-cores 1 $SPARK_HOME/lib/spark-examples*.jar 10

# Spark submit example in cluster mode
spark-submit --class org.apache.spark.examples.SparkPi --master yarn-cluster --num-executors 3 --driver-memory 512m --executor-memory 512m --executor-cores 1 $SPARK_HOME/lib/spark-examples*.jar 10


# Spark shell with yarn client
spark-shell --master yarn-client --num-executors 3 --driver-memory 512m --executor-memory 512m --executor-cores 1


# Pyspark
# -------


# Optionally export Hadoop COnf and PySpark Python
export HADOOP_CONF_DIR=/etc/hadoop/conf
export PYSPARK_PYTHON=/opath/to/bin/python


# PySpark submit example in local mode
spark-submit --verbose /usr/hdp/2.3.0.0-2557/spark/examples/src/main/python/pi.py 100


# PySpark submit example in client mode
spark-submit --verbose --master yarn-client /usr/hdp/2.3.0.0-2557/spark/examples/src/main/python/pi.py 100


# PySpark submit example in cluster mode
spark-submit --verbose --master yarn-cluster /usr/hdp/2.3.0.0-2557/spark/examples/src/main/python/pi.py 100


# PySpark shell with yarn client
pyspark --master yarn-client

@jigar.patel