Stratio/Spark-MongoDB

Spark library for easy MongoDB access

Andrew Kelley

@kelleyaj

Hello i am having some issues following your scala code example, get the error - not found: value createDataFrame. Am i missing an import or something?

Pablo Francisco Pérez Hidalgo

@pfcoperez

Hi @kelleyaj , the example for the Scala API is making use of spark-shell. Hence, the scope has sqlContext reference on it by default.

the line import sqlContext._ brings that SQLContext#createDataFrame to the same scope so it can be called right away.

If you want to easily reproduce the example environment you can download a binary distribution of spark and launch its spark shell using --packages option to include spark-mongodb dependency. e.g:

cd somewhereyouextracted_sparkbins/
./bin/spark-shell --packages com.stratio.datasource:spark-mongodb_2.10:0.11.1

Pablo Francisco Pérez Hidalgo

@pfcoperez

Alternatively, you might create a SQLContextinstance by yourself in your spark application code:

...
val sqlContext = new SQLContext(sc) // or new HiveContext(sc)
import sqlContext._

I hope that helps.

wingerli

@wingerli

Hi, could somebody help me , how to update the collection? I know how to insert and select.

Pablo Francisco Pérez Hidalgo

@pfcoperez

@wingerli You can update by inserting the data within a DataFrame thanks to MongodbDataFrame saveToMongodb method.

Pablo Francisco Pérez Hidalgo

@pfcoperez

Alternatively, you can use Casbah library which is already a dependency of Spark-MongoDB

xuskorea

@xuskorea

Hi i've doubt this framework support batching?

i need to do "lazy" dataframe

Miguel Angel Fernandez Diaz

@mafernandez-stratio

Hi @xuskorea, dataframe are lazy by design, that is, you can add transformation to the dataframe and those transformation are not applied until an output action is invoked

Spark-MongoDB makes batch operations using Spark and accessing efficiently to MongoDB

tunasalat

@tunasalat

Hi! I'm trying to get Spark-MongoDB working. I use the example in the First_Steps document, but get the exception:

# pyspark --packages com.stratio.datasource:spark-mongodb_2.10:0.10.3
>>> from pyspark.sql import SQLContext
>>> sqlContext.sql("CREATE TEMPORARY TABLE col_table USING com.stratio.datasource.mongodb OPTIONS (host 'host:port', database 'db', collection 'col')")
[...]
Exception in thread "Thread-1998" java.lang.NoClassDefFoundError: org/apache/hadoop/hive/ql/parse/VariableSubstitution
    at java.lang.Class.getDeclaredMethods0(Native Method)
    at java.lang.Class.privateGetDeclaredMethods(Class.java:2625)
    at java.lang.Class.privateGetPublicMethods(Class.java:2743)
    at java.lang.Class.getMethods(Class.java:1480)
    at py4j.reflection.ReflectionEngine.getMethodsByNameAndLength(ReflectionEngine.java:365)
    at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:317)
    at py4j.reflection.ReflectionEngine.getMethod(ReflectionEngine.java:342)
    at py4j.Gateway.invoke(Gateway.java:252)
    at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:133)
    at py4j.commands.CallCommand.execute(CallCommand.java:79)
    at py4j.GatewayConnection.run(GatewayConnection.java:207)
    at java.lang.Thread.run(Thread.java:745)

I'm using Spark 1.5.0 from cloudera. I'll provide any other information if needed. Could you point out where I am wrong here?

David Arroyo Cazorla

@darroyocazorla

Hi @tunasalat. It seems to be a known issue (https://community.cloudera.com/t5/Advanced-Analytics-Apache-Spark/I-am-using-a-hive-cotext-in-pyspark-cdh5-3-virtual-box-and-i-get/td-p/24418).

This message was deleted

tunasalat

@tunasalat

@darroyocazorla I've tried it, but no luck. According to log lines at pyspark startup.

2016-05-10 13:59:10,404 WARN  [Thread-2] spark.SparkConf (Logging.scala:logWarning(71)) - Setting 'spark.executor.extraClassPath' to '  [ ... ]   ' as a work-around.

Both spark.executor.extraClassPath and spark.driver.extraClassPath contain jars from /usr/lib/hive/lib/ . Maybe I'm still missing something?

Miguel Angel Fernandez Diaz

@mafernandez-stratio

Hi @tunasalat, as @darroyocazorla mentioned above, this is an issue that is not related to the Spark-MongoDB datasource, you will be facing the same problem is you use any other datasource. Therefore, it's a problem of how Cloudera distribution makes use of Spark so you'll find the proper support for this issue in the Cloudera forums

tunasalat

@tunasalat

@miguel0afd thanks! I'll try my luck there.

Ankit Gohil

@gohilankit

How can I add query filters to load selective data from MongoDB using PySpark?
Using

reader = sqlContext.read.format("com.stratio.datasource.mongodb")
data = reader.options(host='ip:27017', database='db', collection='coll').load()

would load the whole collection into dataframe while I only want to use a chunk of the collection. Is there any equivalent of Mongo's find() method where I can specify query filters?

Miguel Angel Fernandez Diaz

@mafernandez-stratio

Hi @gohilankit

the load method is a lazy one, therefore, you can add transformations (also lazy) to the variable data and they won't be applied until an output action is called

For more information, please, visit https://spark.apache.org/docs/1.6.1/api/python/pyspark.sql.html#pyspark.sql.DataFrame

Zengda

@johnzeng

Hi, all, I have a question about this lib. how do I access to an array inside a document? like , I have a document: { "day":"2016-06-06", "items":[{"ts":123, "val":123}]} , how can I access to items.ts?

Pablo Francisco Pérez Hidalgo

@pfcoperez

Hi @Misfit-John , according to SparkSQL syntax you can try with: SELECT items[0].ts FROM ....

Dipayan

@dipayan90

Hi, I am using Java, Spark and the spark mongo library for one of our applications. I was wondering is there a way I can query using mongo syntax in addtion to sql like statements ... I need to perform an unwind operation on the collection and its simple using a mongo query

Miguel Angel Fernandez Diaz

@mafernandez-stratio

Hi @dipayan90, there is no possibility for using Mongo syntax from our connector. SparkSQL gave birth with the idea of being an SQL abstraction for any data source and unify the way to query the data. Therefore, Spark is not ready for these kind of actions. In addition, there is no other Spark datasource that allows to use native syntax

Dipayan

@dipayan90

@miguel0afd I ended up using spring mongo template to get me the aggregate results I wanted and then converted that into a dataframe ... Anyways thanks for your response... That said https://github.com/mongodb/mongo-hadoop allows you to pass in mongo queries using "mongo.input.query" parameter .... How ever one of their caveats is that they dont allow complex aggregate queries.

jyp

@pzz2011

Execuse me, I wonder whether we can only discuss Spark-MongDB here? Can I ask question about SparkSQL here?

Reinis Vicups

@fahman

Hi, is there a way to eliminate this exception 'CodecConfigurationException: Can't find a codec for class breeze.linalg.SparseVector' without coding my own Codec for breeze Vectors?

Pushpinder Heer

@pheer

What options do i have when using Stratio to access mongo db 2.6.7

Where communities thrive

People

Repo info

Activity