jupyter with Terraform and GCP

February 23, 2023

Table of Contents


This tutorial tells you how to create your customized jupyter environment with GCP. This process is automated by terraform. With this, you don't have to do tedious things at cloud provider's console to setup cloud environments. Terraform is one of IaC(Infrastructure as code) tools. It reduces risk of human error and time for managing it as little as possible.

Maybe you think why you don't use colab or kaggle kernel? Yes, they could be options but there are limitations for available time and resources even though professional service. Thus by setting up your own cloud environment, you can use whatever and as much as you want. But remember you have to pay as you use your resources.


Ubuntu 20.04.4 LTS (Focal Fossa)

Step1 Install terraform

Run this shell if you don't install terraform yet.

sudo apt-get update && sudo apt-get install -y gnupg software-properties-common curl
curl -fsSL https://apt.releases.hashicorp.com/gpg | sudo apt-key add -
sudo apt-add-repository deb [arch=amd64] https://apt.releases.hashicorp.com $(lsb_release -cs) main
sudo apt-get update && sudo apt-get install terraform

Step2 Create VM instance by terraform

Create your service account in GCP for this work, download credential and create terraform file.
Here is an example terraform file.
By enabling preemptible as true you can use preemptible VM. You can take up to 91% discount!
Also, you can set a start-up script. In this example it is set as init.sh.
If you don't use it you can delete the line.

terraform {
  required_providers {
    google = {
      source  = hashicorp/google
      version = 3.5.0

provider google {
  credentials = file(../credential/your_credential.json)
  project     = your_project_id
  region      = asia-northeast1
  zone        = asia-northeast1-a

resource google_compute_instance default {
  name         = test
  machine_type = n1-standard-1

  boot_disk {
    initialize_params {
      image = debian-cloud/debian-10
      size  = 20
  scheduling {
    preemptible       = true
    automatic_restart = false

  metadata_startup_script = file(init.sh)

  metadata = {
    enable-oslogin = TRUE

  network_interface {
    network = default
    access_config {}


Then run this commands to create Google compute Engine instance.
You will find a created vm instances in your GCP console.

$ cd path_to_terraform_file
$ terraform init
$ terraform fmt
$ terraform validate
$ terraform apply

Step3 Set up your instance

Access to a VM instance by ssh then run this shell script to set up Docker container.

curl https://get.docker.com | sh
sudo usermod -aG docker $USER
sudo systemctl start docker
sudo systemctl enable docker
sudo curl -L https://github.com/docker/compose/releases/download/1.16.1/docker-compose-\`uname -s\`-\`uname -m\` -o /usr/local/bin/docker-compose
sudo chmod +x /usr/local/bin/docker-compose
sudo service docker start

# Start jupyter lab
sudo docker-compose up

Here is Dockerfile.

FROM python:3.8.12-buster

USER root

RUN pip install --no-cache-dir \\  

Here is a yaml file for docker compose.

version: 3
    build: ./docker_images/jupyter
      - .:/home/work
      - .jupyter:/root/.jupyter
      - 7777:7777
    tty: true
    command: jupyter lab --ip= --port=7777 --allow-root --no-browser

In this example Dockerfile is set in docker_images/jupyter directory.

Step4 Access to jupyterlab hosted on vm instance

You can do port forwarding to get access to your vm instance like this.

$ gcloud compute ssh --project=[project id] --zone=[zone] [instance name] -- -L [local port#]:[instance ip adress]:7777

Then, you can use jupyter lab by accessing http://localhost:7777/

Step5 Clean up

To delete gcp resources, run this script.

$ terraform destroy

Please share it if you like!

Profile picture

Written by mtzk who lives and works as a programmer in Tokyo.