This repository contains the configuration for a virtual Small Size League tournament with multiple fields and all components that are required.
Following software is required, which should be available on all major operating systems:
As a team, you need to make sure that your software can run on Ubuntu 20.04 and without root permissions. If you require a GPU or have other specific requirements, ask the technical committee if it will be possible.
Each team has its own docker container including a simple desktop environment. If you need additional libraries to run and/or build your software, create a custom docker image based on src/ubuntu-vnc. Use src/ubuntu-vnc-java as a template. Then submit a pull-request with the new docker image. It is possible to copy files into the container, so you can also build your software locally and only copy the binaries over.
Your container will have a volume mounted to the home folder. The volume will be used for all fields, while there will be individual containers per field. You do not need to worry about changing any network addresses or ports, they will always be the same. You should be able to handle multiple network interfaces, though.
To integrate your own container, add a new service to docker-compose-teams.yaml. Take the tigers or erforce service as a template. Also, add your team name to config/teams. It should be all-lower-case without any special characters.
If you want to use a remote server, change the root domain and field name first in ./config/root_domain and ./config/field_name. Else, the defaults are fine.
Before you start anything, you need to initialize some secrets with:
./config/docker/init.sh
This will generate passwords and an SSH key and put them at the right places.
Next, generate the initial Caddyfile for the webserver:
./config/caddy/generate_caddyfile.py
Now, you can spin up the field:
# Start all containers and keep showing the log in the foreground (ctrl+c will stop everything again)
docker compose up
# Or alternatively run all containers in the background:
docker compose up -d
Next, setup Guacamole and caddy:
# Sets up the running Guacamole (VNC) server. Will also generate team passwords.
./config/guacamole/update_guacamole.py
# Convert all passwords into a caddy format for the game-controller
./config/caddy/update_caddy_passwords.sh
# Regenerate the Caddyfile with the newly created passwords
./config/caddy/generate_caddyfile.py
# Reload the running caddy webserver
./config/caddy/update_caddy_config.sh
Finally, spin up the team containers that you need:
# Start team containers (individually or all together)
docker compose -f docker-compose-teams.yaml up [team-container]
If you want to have additional monitoring, you can also start the monitoring containers:
# Start monitoring containers
docker compose -f docker-compose-monitoring.yaml up
You can access the field through your browser now. The URL is https://localhost or whatever you chose as the root domain.
If you are running locally, you might want to add the local CA from caddy to your system with:
# Note: Have a look at this script for more details
./caddy/install_local_CA.sh
All credentials were generated to ./config/passwords.
Stop and remove all containers, networks and volumes (-v
) for a specific field:
docker compose down -v
The full setup can be deployed to AWS with terraform:
cd terraform
terraform init
terraform apply
Make sure to make yourself familiar with AWS and terraform, before doing this. Most notably:
You can connect to the instances with:
ssh terraform@<public-ip>
# Check the cloud-init log
less /var/log/cloud-init-output.log
# Go to setup folder
cd ssl-simulation-setup
# Start team containers
docker compose -f docker-compose-teams.yaml up -d [team-container]
To destroy the setup, run:
terraform destroy
The database setup scripts are pre-generated already, but if the guacamole version changes, the script might need to be generated again with:
docker run --rm guacamole/guacamole /opt/guacamole/bin/initdb.sh --postgresql > init/postgres/01_initdb.sql