Distributed k-means with PySpark

README for internal use.

    ____                                   __   __         
   / __ \____ _      _____  ________  ____/ /  / /_  __  __
  / /_/ / __ \ | /| / / _ \/ ___/ _ \/ __  /  / __ \/ / / /
 / ____/ /_/ / |/ |/ /  __/ /  /  __/ /_/ /  / /_/ / /_/ / 
/_/    \____/|__/|__/\___/_/   \___/\__,_/  /_.___/\__, /  
   __________  ____  __  ______     _   _______   /____/__ 
  / ____/ __ \/ __ \/ / / / __ \   / | / /  _/ | / / ____/ 
 / / __/ /_/ / / / / / / / /_/ /  /  |/ // //  |/ / __/    
/ /_/ / _, _/ /_/ / /_/ / ____/  / /|  // // /|  / /___    
\____/_/ |_|\____/\____/_/      /_/ |_/___/_/ |_/_____/

The CloudVeneto Experience

1. Open a terminal and enter the CV gate machine:

   ssh [email protected]
   

   When prompted, enter the longer password.

2. When inside, access the VM of your choice – usually the master – with

   ssh -i private/gbordin.pem [email protected]   
   

   Use the shorter password this time. If for some reason you want to access the slaves right away, just substitute their IP address on the right of the @. The command prompt should change to

   bordin@mapd-b-2023-gr04-1:~$   
   

   or gr04-2/gr04-3 if you accessed the slaves.

3. Start the cluster by running the script start_cluster.

4. Enter the command sparkup: it should open a jupyter-notebook session on the 9000 port of the VM.

5. Open a new terminal and re-enter the cloud by skipping right to the VM, defining also the tunnelling ports:

   ssh -J [email protected] \
       -L 8008:localhost:9000 \
       -L 1234:localhost:8080 \
       -L 4321:localhost:4040 \
       [email protected]
   

   Substitute 8008, 1234 and 4321 with the ports of your choice. 8008 is for the Jupyter notebook, 1234 for the Spark master page and 4321 for the workers’ dashboard.

6. Open localhost:8008 (1234, 4321, or the ports that you’ve chosen) on your browser and you should be able to access the Jupyter session and the Spark stuff.

7. When you finish your work, remember to stop the cluster by running the stop_cluster script from the master VM.

You can jump between the VMs when you’re inside with ssh master, ssh slave01 or ssh slave02. This is useful when you have to install new Python packages, because you have to do it on all three machines.