how to query data from mongodb with spark in zeeplin?

[#1]

Hello , I recently installed hdp-2. with hdp2.6 sandbox in vmawre ! I runned some job with spark to transform csv data and I saved them in mongodb Now I want to visualise some charts dashboard from my Mongodb data base so I added mongodb interpreter to zeeplin but it seems mongodb not good since I have a collection that contain 2 GB so I decide to work with spark ! how can I read data from Mongodb ! I must import some library for example com.mongodb.spark.sql._ and com.mongodb.spark.config.ReadConfig com.mongodb.spark.MongoSpark how I can do this in zeeplin !

in addition spark interpreter is better than mongodb interpreter in this context.

Thanks in advance

[#2]

@chaouki trabelsi

any updates on this issue ? it is very important.

[#3 · Accepted solution]

@chaouki trabelsi @Victor

There are 2 approaches you can take. One is using package and the other is using jars (you need to download the jars)

Package approach

Add the following configuration on your zeppelin spark interpreter:

spark.jars.packages = org.mongodb.spark:mongo-spark-connector_2.11:2.2.2
# for more information read here https://spark-packages.org/package/mongodb/mongo-spark

Jar approach

You need to add the mongo db connector jars to the spark interpreter configuration.

1. Download the mongodb connector jar for spark (depending on your spark version make sure you download the correct scala version - if spark2 you should use 2.11 scala)

2. Add the jars to the zeppelin spark interpreter using spark.jars property

spark.jars = /location/of/jars

On both cases you need to save and restart the interpreter.

HTH

*** If you found this answer addressed your question, please take a moment to login and click the "accept" link on the answer.

[#4]

@Felix Albani Can you kindly help me with this error

And my code is

%spark2.pyspark
from pyspark.sql import SparkSession
my_spark = SparkSession     .builder     .appName("myApp")     .config("spark.mongodb.input.uri", "mongodb://127.0.0.1/db.col")     .config("spark.mongodb.output.uri", "mongodb://127.0.0.1/db.col")     .getOrCreate()

And the output is

<pyspark.sql.session.SparkSession object at 0x7ffa96a92c18> 

Then this one causes an error

df = my_spark.read.format("com.mongodb.spark.sql.DefaultSource").load()

And the error is

": java.lang.NoClassDefFoundError: com/mongodb/ConnectionString"

Here is the jar file I added in the Zeppelin interpreter

[#5]

Thanks @Felix Albani ! I have a question! is spark interpreter is the best in this case? is spark work with mongodb in the same way with hdfs (memory+speed) ??

[#6]

@chaouki trabelsi mongodb connector is build to leverage spark parallelism. So I think is a good alternative on this case. If you have further questions on how to use it or anything else please open a separate thread! Thanks!

[#7]

guys I am having the following issue trying to query mongo db from zeppelin with spark:

java.lang.IllegalArgumentException: Missing collection name. Set via the 'spark.mongodb.input.uri' or 'spark.mongodb.input.collection' property

I have set mongo-spark-connector_2.11:2.2.2 in dependencies of spark2 interpretator

and my code is:

%spark2
import com.mongodb.spark._
spark.conf.set("spark.mongodb.input.uri", "mongodb://myip:myport/mydb.collection")
spark.conf.set("spark.mongodb.output.uri", "mongodb://myip:myport/mydb.collection")

val rdd = MongoSpark.load(sc)

I also tried:

%spark2
sc.stop()
import org.apache.spark.sql.SparkSession
import com.mongodb.spark._
import com.mongodb.spark.config._

val spark_custom_session = SparkSession.builder()
      .master("local")
      .appName("ZeplinMongo")
      .config("spark.mongodb.input.database", "mongodb://myip:myport/mydb.collection")
      .config("spark.mongodb.output.uri", "mongodb://myip:myport/mydb.collection")
      .config("spark.mongodb.output.collection", "mongodb://myip:myport/mydb.collection")      .getOrCreate()
val customRdd = MongoSpark.load(spark_custom_session)
rdd.count

And

import com.mongodb.spark.config._
val readConfig = ReadConfig(Map(
    "spark.mongodb.input.uri" -> "mongodb://myip:myport/mydb.collection", 
    "spark.mongodb.input.readPreference.name" -> "secondaryPreferred"), 
    Some(ReadConfig(sc)))
val customRdd = MongoSpark.load(sc, readConfig)
customRdd.count

What ever I do I get:

import org.apache.spark.sql.SparkSession import com.mongodb.spark._ import com.mongodb.spark.config._ spark_custom_session: org.apache.spark.sql.SparkSession = org.apache.spark.sql.SparkSession@4f9c7e5f java.lang.IllegalArgumentException: Missing collection name. Set via the 'spark.mongodb.input.uri' or 'spark.mongodb.input.collection' property at com.mongodb.spark.config.MongoCompanionConfig$class.collectionName(MongoCompanionConfig.scala:270) at com.mongodb.spark.config.ReadConfig$.collectionName(ReadConfig.scala:39) at com.mongodb.spark.config.ReadConfig$.apply(ReadConfig.scala:60) at com.mongodb.spark.config.ReadConfig$.apply(ReadConfig.scala:39) at com.mongodb.spark.config.MongoCompanionConfig$class.apply(MongoCompanionConfig.scala:124) at com.mongodb.spark.config.ReadConfig$.apply(ReadConfig.scala:39) at com.mongodb.spark.config.MongoCompanionConfig$class.apply(MongoCompanionConfig.scala:113) at com.mongodb.spark.config.ReadConfig$.apply(ReadConfig.scala:39) at com.mongodb.spark.MongoSpark$Builder.build(MongoSpark.scala:231) at com.mongodb.spark.MongoSpark$.load(MongoSpark.scala:84) ... 73 elided

PLEASE HELP! 🙂

[#8]

Hello @Daniel Pevni what's spark version and mongodb version ?

[#9]

spark2 mongoldb 3.2