This project can be used to spin-up a droplet (virtual machine) on Digital Ocean.

This droplet is configured on first boot in order to provide various services to end users.

The droplet OS is Ubuntu 20.04 (focal fossa).

• A default user named lxx is created with home directory /home/lxx
• This user can use sudo without password and is a member of docker group

journald log system is configured to store logs persistently by default.

• Password authentication is disabled
• SSH keypairs must be used
• Port 22 is used by default
• A single SSH key can connect to both root and lxx user by default

Note: Use the DO_SSH_KEY_NAME environment variable to configure the name of SSH key to allow.

• Time synchronisation is enabled using systemd-timesyncd.
• The default NTP server is time.windows.com.

• Docker Engine is installed and can be used using the docker command.
• Only users in docker group can run docker commands without sudo
• Engine configuration can be found under /etc/docker/daemon.json

• A single-node docker swarm cluster is running by default.
• Docker Swarm cluster advertise address is the droplet private address within its VPC (it means that it's possible to create a new droplet in the same VPC and join the swarm cluster)

• Traefik reverse proxy is installed using Docker Swarm and listens on ports 80 and 443 on host machine.
• Traefik dashboad can be reached on https://traefik.<domain> (default https://traefik.lxx.quest).

Note: Update DOMAIN_NAME to configure domain to use. For example, using DOMAIN_NAME=test.lxx.quest leads to Traefik Dashboard listenning on https://traefik.test.lxx.quest

In order to expose docker services using Traefik, containers must be attached to the traefik-network (docker overlay network).

• MinIO server is installed using Docker Swarm and is exposed behind Traefik:
  • MinIO Dashboard can be reached on https://minio.<domain> (default https://minio.lxx.quest)
  • MinIO Server can be reached on https://storage.<domain> (default https://storage.lxx.quest)

• Jupyterhub is installed using TLJH
• Jupyter users are authenticated using systemd.
  • Admin user is lxx
    • Password is defined on first connection
  • New users can be created from admin panel
    • Unix users will be created automatically

• TLS certificates are available under /etc/lego/certificates
• Certificates are renewed 30 days before expiration (check performed daily)

• A Linux system is required to contribute to the project.

• A Linux system is required to use the project.

Note: terraform can be used on Windows, as well as task, but the Taskfile is not cross-platform.

• Install Terraform - https://learn.hashicorp.com/tutorials/terraform/install-cli#install-terraform

• Install Task - https://taskfile.dev/installation/

• Install doctl - https://docs.digitalocean.com/reference/doctl/how-to/install/

• Install terraform-docs

• List droplet sizes:

doctl compute size list --output json | jq -r "[.[].slug]"

• List droplet regions:

doctl compute region list --output=json | jq '[.[].slug]'

• Make sure terraform is installed.

• Make sure task is installed

• Create a file named .env and write options:

DO_TOKEN: <secret>
DO_SSH_KEY_NAME: <your_ssh_key_name>
DO_DOMAIN_NAME: lxx.quest
DOMAIN_NAME: test.lxx.quest
DO_SSH_KEY_NAME: charbonnierg@WSL2
DROPLET_SIZE: s-2vcpu-4gb
DROPLET_NAME: lxx-test
DB_CLUSTER_NAME: lxx-test-cluster
DB_CLUSTER_SIZE: db-s-1vcpu-1gb
DO_REGION: ams3

Don't forget to update and review the values. Cost will depend on configured sizes.

The configuration used above has a total cost of 35$ / month:

• droplet: 20$ / month
• database cluster: 15$ / month

Other resources do not have a cost (domain records and firewalls)

• Initialize terraform environment:

task init

• Prepare deployment

task plan

You should see a summary of resources to create in the console:

- Apply deployment

```bash
task apply

• Inspect the droplet state:

task show-droplet

• Connect to the droplet:

task ssh

• Connect using VSCode:

task code

When using WSL, Windows user SSH key must be allowed to connect to remote host instead of Linux user SSH key

Navigate into the tf/ directory and use terraform CLI directly to:

• provide variables
• plan deployment
• apply deployment
• destroy deployment