Install Snowpark Submit¶

You can install Snowpark Submit to run batch-oriented Spark workloads directly on Snowflake’s infrastructure.

To install Snowpark Submit, complete the following steps:

1. Install Snowpark Submit by using pip.

   pip install snowpark-submit
   

   Copy

2. In a connections.toml file for Snowflake authentication, add a connection named snowpark-submit.

   If you don’t have such a file already, create one as described in Connecting using the connections.toml file.

   [snowpark-submit]
   host = "<account>.snowflakecomputing.cn"
   port = 443
   account = "<account>"
   user = "test_user"
   role = "test_role"
   password = "<password for user>"
   protocol = "https"
   warehouse = "test_warehouse"
   database = "test_db"
   schema = "test_schema"
   compute_pool = "test_compute_pool"
   

   Copy

3. Verify that you can connect to Snowflake from your client computer.

   To verify that the connection works from your client computer, create a .py file with code that connects to Snowflake.

   1. Create a connection_test.py file, and then add the following code:

      # connection_test.py code
      
      import sys
      import snowflake.connector
      
      conn_name = sys.argv[1]
      
      print(f"Trying connection named {conn_name}..")
      conn = snowflake.connector.connect(connection_name=conn_name)
      print("Connected.")
      
      cursor = conn.cursor()
      cursor.execute("SELECT 'Connection successful'")
      for col in cursor:
          print(col)
      
      print("\nListing first 5 tables:\n")
      cursor = conn.cursor()
      cursor.execute('show tables limit 5')
      for col in cursor:
          print(col)
      print("\nDone")
      

      Copy

   2. From your active Python virtual environment, run the following command, specifying the name of the connection that you added to your connections.toml file.

      python connection_test.py snowpark-submit
      

      Copy