Deploy JupyterHub for AI Training on Droplets.

Question

Pulumi · Accepted Answer

To deploy JupyterHub for AI training on DigitalOcean Droplets using Pulumi, you need to create the infrastructure that consists of the Droplets (virtual machines) where the JupyterHub will run. You will also need a mechanism to execute a startup script that sets up JupyterHub on your Droplet after it's created.

Here's a Pulumi Python program that demonstrates how to create a DigitalOcean Droplet and run a startup script to install JupyterHub:

```python
import pulumi
import pulumi_digitalocean as digitalocean

# Configuration for the Droplet
droplet_name = 'ai-training-droplet'
region = 'nyc3' # New York 3 datacenter
size = 's-2vcpu-4gb' # Standard Droplet with 2 vCPUs and 4GB RAM
image = 'docker-20-04' # Slug for a Docker pre-installed image

# User data (startup script) to install and setup JupyterHub
user_data = '''#!/bin/bash
apt-get update
apt-get -y install python3 python3-pip
pip3 install jupyterhub notebook

# Create a JupyterHub user - you can replace 'jupyterhub_user' with your desired user name
useradd -m jupyterhub_user
echo jupyterhub_user:password | chpasswd

# Run JupyterHub on startup
nohup jupyterhub --port 80 &
'''

# Create a new SSH key to use with the Droplet
ssh_key = digitalocean.SshKey("example_ssh_key",
    name="example_ssh_key",
    public_key="ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQC...")

# Instantiate a Droplet
droplet = digitalocean.Droplet(droplet_name,
                               region=region,
                               size=size,
                               image=image,
                               user_data=user_data,
                               ssh_keys=[ssh_key.id],
                               tags=["jupyterhub", "training"])

# Export the Droplet's IP address
pulumi.export('droplet_ip', droplet.ipv4_address)
```

In the above code:

- We import the `pulumi` and `pulumi_digitalocean` modules to interact with Pulumi and DigitalOcean respectively.
- We define configurations for our Droplet such as the name, region, size, image, and a startup script `user_data`. The startup script is executed when the Droplet is created, installing Python, JupyterHub, and creating a JupyterHub user account.
- An SSH key is created for secure access to the Droplet.
- The DigitalOcean Droplet is provisioned with the specified configurations. We attach the SSH key and assign it tags for organization.
- Lastly, we export the IP address of the Droplet so that it can be accessed once it's ready.

To run this program:

1. Ensure you have Pulumi installed and configured with DigitalOcean.
2. Save the above code in a file named `__main__.py`.
3. Run `pulumi up` in the same directory to create the cloud resources.

Remember to secure your JupyterHub installation properly by setting up a more secure authentication method and HTTPS before using it for sensitive workloads. This example uses a basic setup that should not be used as-is for a production environment.