# Spark Docker for centos7


## Running a Standalone Spark cluster using Docker-compose
1. Build the image ```docker-compose build```
2. Start up a Spark master and n slaves ```docker-compose up -d && docker-compose scale slave=n```; change n with the number of required slaves.
3. check the created network by compose ```docker network ls```. This network can be adjusted in the docker-compose.yml file. if you are using vagrant then by default a "vagrant" prefix will be created e.g "vagrant_spark_network"
4. to destroy the cluster ```docker-compose down```


## Run interactive shell
docker run -it --net spark_network nhjiejan/spark /opt/spark/bin/spark-shell --master spark://master:7077

## Run spark submit
you can specify additional volume mount points, the mount created exists in "/vagrant/spark_data". place your fat jar in this directory and run:

```docker run -it --net spark_network nhjiejan/spark /opt/spark/bin/spark-submit --master spark://master:7077 ...```

you should specify the above "--master" as this will connect to the slave nodes.