mozilla / jupyter-spark

Jupyter Notebook extension for Apache Spark integration
Mozilla Public License 2.0
193 stars 34 forks source link

Spark/YARN Setup: --deploy-mode client --master yarn / HTTPS Endpoint #43

Open mafeko opened 6 years ago

mafeko commented 6 years ago

Hi your tools looks really promising.

currently i get: {"error": "SPARK_NOT_RUNNING"} running /spark in my setup.

How can i configure the extension to work in an environment like this: pyspark2 --deploy-mode client --master yarn

Could you give me a hint how to configure the setting in this case? jupyter notebook --Spark.url="http://localhost:4040"

Thank you.

mafeko commented 6 years ago

I think I got a little closer to the solution,

I could find out the correct spark-endpoint in my setup by looking up the configuration of the spark-context (sc):

print(sc.getConf().get('spark.org.apache.hadoop.yarn.server.webproxy.amfilter.AmIpFilter.param.PROXY_URI_BASES').split(','))

In my case this is a https url, which now seems to be another challenge.

lucasdurand commented 5 years ago

Hey, any more on this? I'm looking at rigging this up with Yarn as well