# PySpark

Steps:

* bin/pyspark call org.apache.spark.launcher.Main which would build the command to launch python. And also set the shell.py as the startup script of pyspark to start the spark. 
* context.py will call java\_gateway.py to launch spark by invoking bin/spark-submit. 
* spark-submit will launch JavaGatewayServer
* Python side create SparkConf and SparkContext through the gateway

bin/pyspark.sh

```
export PYTHONSTARTUP="${SPARK_HOME}/python/pyspark/shell.py"
```