Containers

Release tags publish an image to ghcr.io/brickadia-community/omegga.

docker run -it --rm \
  -v omegga-home:/home/steam \
  -v "$PWD/server:/server" \
  -p 8080:8080 -p 7777:7777/udp \
  -e BRICKADIA_TOKEN -e OMEGGA_PORT=8080 -e BRICKADIA_PORT=7777 \
  -e PUID="$(id -u)" -e PGID="$(id -g)" \
  ghcr.io/brickadia-community/omegga

Generate a hosting token at https://brickadia.com/account/tokens; without one omegga stops at an interactive auth prompt on first start. Arguments are passed through, so omegga --debug and bash both work in place of the default command. Omegga handles SIGINT and SIGTERM, so ctrl+c and docker stop shut down gracefully - add --init if you also want zombie reaping.

Omegga is baked into the image, so it updates by pulling a new one rather than through npm:

docker pull ghcr.io/brickadia-community/omegga:latest
docker compose pull && docker compose up -d # or, under compose

latest moves with every release, so pin a version like :1.14.0 if you would rather choose when that happens. Nothing pulls on its own - podman auto-update and watchtower are the usual ways to automate it.

The game is not in the image. It updates in place inside the volume with omegga --update or /update, and steamcmd keeps itself up to date. Setting the container’s command to omegga --update (command: omegga --update under compose) checks on every start, at the cost of not starting at all when Steam is unreachable, rather than running the version already installed.

Compose

services:
  omegga:
    image: ghcr.io/brickadia-community/omegga:latest
    restart: unless-stopped
    stdin_open: true
    tty: true
    env_file: .env
    ports:
      - '${OMEGGA_PORT}:${OMEGGA_PORT}/tcp'
      - '${BRICKADIA_PORT}:${BRICKADIA_PORT}/udp'
    volumes:
      - home:/home/steam
      - ./server:/server

volumes:
  home:

# .env - read for the ${...} above, and passed to the container
BRICKADIA_TOKEN=...
OMEGGA_PORT=8080
BRICKADIA_PORT=7777
PUID=1000
PGID=1000

compose run --rm --service-ports omegga attaches a tty, so ctrl+c goes to omegga. With compose up it goes to compose, which stops the container instead.

Volumes

/home/steam - the Brickadia install, steamcmd, and auth files. Keep it on a named volume or the game is downloaded again every time the container is recreated.
/server - the omegga working directory: omegga-config.yml, data, and plugins. Bind mount it to edit them from the host.

Ports

Set them with OMEGGA_PORT and BRICKADIA_PORT rather than editing the config file, so one value drives both the server and the port mapping. They override the config, though server.port still has to be present in it. See Environment Variables for the rest.

Brickadia reports its configured port to the master server, so a published port has to be the same number the server runs on. Never map -p 7778:7777/udp.

Metrics

METRICS_ENABLED=true with METRICS_BIND=0.0.0.0 serves the Prometheus endpoint on the container’s own network interface, where a scraper on the same compose network reaches it by service name without the port being published at all. metrics.token has no environment variable, so it belongs in omegga-config.yml on the mounted /server.

Metrics has a compose file with omegga, Prometheus, and VictoriaMetrics together, and the config files that go with it.

File Ownership

PUID and PGID decide who owns what omegga writes into a bind-mounted /server; set them to your own id -u and id -g. PUID=0 runs as root instead and leaves root-owned files behind on the host.

Building

The Dockerfile builds omegga on top of gameservermanagers/steamcmd, with node from nvm.

# the omegga in this checkout
docker build -t omegga .

# or a release off npm, where OMEGGA_VERSION is an npm version
docker build -t omegga --target npm --build-arg OMEGGA_VERSION=1.14.0 .

OMEGGA_VERSION is required and has to be an exact version. There is deliberately no latest default: that layer is cached on its command text, so latest would never invalidate it - rebuilds would keep reinstalling whichever version was current the first time, and only --no-cache would get past it.

Podman

podman build, podman run, and podman compose take the same arguments and read the same compose file. Two things differ:

Rootless podman maps the container’s root to your own user, so PUID=0 is what leaves bind-mounted files owned by you - the opposite of the advice above. Keeping the container’s unprivileged user maps it to a subuid instead, and /server ends up owned by an id you need podman unshare to touch. To stay unprivileged inside the container, --userns=keep-id:uid=1000,gid=1000 maps you onto the image’s steam user.
On SELinux systems bind mounts need a relabel: -v "$PWD/server:/server:Z".

Keyboard shortcuts

Omegga