Reading and writing to minIO from Spark

minio_spark.py

from pyspark import SparkContext, SparkConf, SQLContext
import os

os.environ['HADOOP_HOME'] = '/opt/hadoop/'
os.environ['JAVA_HOME'] = '/usr/lib/jvm/java-8-openjdk-amd64'
os.environ['PYSPARK_DRIVER_PYTHON'] = 'python3'
os.environ['PYSPARK_PYTHON'] = 'python3'
os.environ['LD_LIBRARY_PATH'] = '/opt/hadoop/lib/native'
os.environ['SPARK_DIST_CLASSPATH'] = "/opt/hadoop/etc/hadoop:/opt/hadoop/share/hadoop/common/lib/*:/opt/hadoop/share/hadoop/common/*:/opt/hadoop/share/hadoop/hdfs:/opt/hadoop/share/hadoop/hdfs/lib/*:/opt/hadoop/share/hadoop/hdfs/*:/opt/hadoop/share/hadoop/mapreduce/lib/*:/opt/hadoop/share/hadoop/mapreduce/*:/opt/hadoop/share/hadoop/yarn:/opt/hadoop/share/hadoop/yarn/lib/*:/opt/hadoop/share/hadoop/yarn/*"
os.environ['SPARK_HOME'] = '/opt/spark/'

conf = (
    SparkConf()
    .setAppName("Spark Minio Test")
    .set("spark.hadoop.fs.s3a.endpoint", "http://localhost:9091")
    .set("spark.hadoop.fs.s3a.access.key", os.environ.get('MINIO_ACCESS_KEY'))
    .set("spark.hadoop.fs.s3a.secret.key", os.environ.get('MINIO_SECRET_KEY'))
    .set("spark.hadoop.fs.s3a.path.style.access", True)
    .set("spark.hadoop.fs.s3a.impl", "org.apache.hadoop.fs.s3a.S3AFileSystem")
)
sc = SparkContext(conf=conf).getOrCreate()
sqlContext = SQLContext(sc)

# Reading
print(sc.wholeTextFiles('s3a://datalake/test.txt').collect())

# Writing
path = "s3a://user-jitsejan/mario-colors-two/"
rdd = sc.parallelize([('Mario', 'Red'), ('Luigi', 'Green'), ('Princess', 'Pink')])
rdd.toDF(['name', 'color']).write.csv(path)

@jitsejan I need help in setting up PySpark and access to AWS S3

What have you done so far? S3 will work slightly different than Minio. Take a look at an older post I did: https://www.jitsejan.com/using-spark-to-read-from-s3.html.

I have a Dockerfile with Base image jupyter/scipy-notebook and then downloading Spark with Hadoop, sylinking jars and then maintaining PATH and PYTHON_PATH and SPARK_HOME. Downloading aws.sdk and hadoop-aws-sdk. I am gettign Client exception in the decoding process. I am trying to read a json into dataframe i.e. Py4j error is raised.

I am using this Dockerfile https://github.com/jupyter/docker-stacks/blob/master/pyspark-notebook/Dockerfile

I have tried your post but still no luck. I get following exception during reading from s3 using spark.
When jupyter/pyspark-notebook is used as base image and then installed aws sdk and hadoop aws sdk additionally.

When jupyter/scipy-notebook is used as base image and then installed aws sdk and hadoop aws sdk additionally.

When jupyter/scipy-notebook is used as base image and then installed aws sdk and hadoop aws sdk additionally.

Did you use those exact same versions of the libraries? I remember I struggled using different SDK versions and only this combination worked for me at the time.

I followed the instructions from your blog and having trouble in importing pyspark

That's resolved and now I get this new error. It seems to use spark from pyspark-notebook image i..e from /usr/local/spark although SPARKHOME is set to path from "/home/jovyan/spark-2.4.0-bin-hadoop2.7"

That's resolved and now I get this new error. It seems to use spark from pyspark-notebook image i..e from /usr/local/spark although SPARKHOME is set to path from "/home/jovyan/spark-2.4.0-bin-hadoop2.7"

Could you try to set SPARK_HOME explicitly in your notebook so you are sure it is the right version? Did you also try with findspark?