ETC5513: Collaborative and Reproducible Practices

Workshop 11

Workshop Goals

By the end of this workshop, you’ll:

1. Have Docker running on your own machine
2. Run RStudio Server in a container
3. Launch your own local chatbot using ollama and open-webui
4. Understand how to use Docker Compose to manage complex environments

---

Part 1: Set up Docker

1. Install Docker Desktop

• Go to https://www.docker.com/products/docker-desktop
• Download Docker Desktop for your operating system (Mac, Windows, or Linux)
• Follow the installation instructions
• Launch Docker Desktop

To test your setup, run this in the terminal:

docker run hello-world

If you see a “Hello from Docker!” message, you’re good to go!

---

Part 2: Run RStudio Server with Docker Compose

You’ll use a Docker Compose file to spin up a full RStudio Server container.

1. Create a new folder for this project and add this file as docker-compose.yml:

services:
  rstudio:
    image: rocker/rstudio
    ports:
      - "8787:8787"
    environment:
      - PASSWORD=rstudio
    volumes:
      - .:/home/rstudio
    container_name: rstudio-docker

2. From the terminal in that folder, run:

docker compose up -d

3. Open your browser and go to:

http://localhost:8787

• Username: rstudio
• Password: rstudio

🧪 Try making a small script or loading a package!

To stop the container:

docker compose down

---

Part 3: Run a Local Chatbot with Open WebUI + Ollama (Bundled)

In this section, we will run a local chatbot using ollama and open-webui.

1. Create a new folder and inside it, add this docker-compose.yml:

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:ollama
    ports:
      - 3000:8080
    volumes:
      - open-webui-data:/app/backend/data

volumes:
  open-webui-data:

This setup includes Ollama already inside the container and will download models as needed.

---

2. Run the chatbot:

From the terminal in that folder:

docker compose up -d

Once it finishes setting up, visit:

http://localhost:3000

Set up a local account, and select model llama3.2. This will take some time to download!

---

3. Try it out!

• Ask the model some questions.
• Open multiple tabs to simulate multiple sessions.
• Try using larger models like llava or codellama if you’re curious (and have RAM to spare).

To stop the server:

docker compose down

---

Note: This bundled version is great for local experimentation, and includes both the LLM and a clean interface. It is entirely possible to run ollama separately from open-webui:

You do not need to do this step in the workshop

1. Install Ollama

• Go to https://ollama.com and install for your OS

2. Run ollama in the background:

ollama serve

Then pull a model like:

ollama run llama2

Keep this running!

3. Set up open-webui with Docker Compose

Create another folder and add this docker-compose.yml:

services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    ports:
      - 3000:3000
    volumes:
      - open-webui-data:/app/backend/data
    environment:
      - OLLAMA_BASE_URL=http://host.docker.internal:11434

volumes:
  open-webui-data:

Then run:

docker compose up -d

Visit: http://localhost:3000