Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?


Failed to load latest commit information.
Latest commit message
Commit time

Putting game servers in Docker containers. Because it makes life easier.

A set of game servers that have been wrapped with Docker. All images are composed of other images in this repository, except for base. Most images will auto-download any dependencies they have (eg, factorio server files or sourcemod for srcds), making it almost one command to install a server. No binaries are stored in this repo, so make sure you have internet to download them with.

Includes a utility script which mounts save data to an external volume and sorts out networking for you.

Quick start

Assuming you are on a blank server with Git and docker installed:

git clone .
./ factorio

./ factorio
# Or
docker run -it --net=host -e "SERVER_NAME=Some really cool server"\
-v `pwd`/save:`cat factorio/mounts` --name factorio factorio

Most servers have environment variables that can be used to configure them. Read the file in each directory to see what is available.


There is a tutorial on the Open Source LAN website which will step you through building and using the images.


Contributions are welcome. Submit a pull request or open an issue.

All additions should follow these suggestions:

  • Set up for LAN by default - eg, call +sv_lan 1 in srcds games, but optionally disabling that by runtime var is fine
  • Include a start script that is added to the container, so anyone using the Dockerfile can easily customise their server

Advanced networking stuff

We want to expose all game servers directly to our LAN, and the --host option means we will get port conflicts. There is an alternative approach which uses the ipvlan or macvlan docker network driver.

# Creates a docker network that's bridge with your layer 2 network
# Subnet should match the IP range and subnet of your network.
# ip-range is the CIDR block of IP addresses to assign to containers
# parent is the name of the interface you'd like to bridge containers to

docker network create --driver ipvlan --subnet= -o parent=eth0 --ip-range --gateway  gameservers

# If the above command doesn't work, try using macvlan instead of ipvlan

# Starts your game server inside the layer 2 network
docker run -it --rm --net=gameservers csgo /steam/csgo/srcds_run -game csgo +sv_lan 1 +map cs_office

You can now see the CSGO server from another server on your network.

Running game servers on Docker Swarm

These are the instructions for building a swarm.

In addition to the swarm, one also needs a docker registry (I think).

These are the following gotchas that you need to be aware of:

  • If using ipvlan, you need experimental, not main release docker (as at version 1.12)
  • The ipvlan network is local scoped. This means you need to create the network on every host individually. You should ensure the assigned IP ranges for each host do not overlap, as there does not appear to be any shared state about IP addresses assigned. The name of the networks should be identical on every host.
  • The tutorial only has you set up a single consul instance. It will change IP addresses every time the container restarts, and this will cause consul to fail because it cannot elect itself leader, since it thinks another instance exists at a different IP address. Make sure to hard code an IP for it.
  • Using ubuntu 14.04, you'll need to update /etc/defaults/docker to contain: DOCKER_OPTS="-H tcp:// -H unix:///var/run/docker.sock"
  • If not using a secured docker registry, also add to the DOCKER_OPTS value: --insecure-registry=registry-hostname:5000
  • Consul, docker node and docker masters can all be on the same host (I think - it works for me at least)
  • To get a container to failover between hosts in the case of an outage, you need to run the container with these args: -e reschedule:on-node-failure (todo: figure out why the container doesn't get auto-started on the new host). Also note that a race condition exists - when the host comes back up, it will start the old container, and when the docker swarm manager container starts, it will detect the container was moved and kill it.

Setting up multiple consul hosts

Run the master like this:

docker run -d  -v /mnt:/data --name consul \
    -p 8300:8300 \
    -p 8301:8301 \
    -p 8301:8301/udp \
    -p 8302:8302 \
    -p 8302:8302/udp \
    -p 8400:8400 \
    -p 8500:8500 \
    -p 53:53/udp \
    progrium/consul -server -advertise -bootstrap-expect 3

and the other two like this:

    docker run -d -v /mnt:/data --name consul \
    -p 8300:8300 \
    -p 8301:8301 \
    -p 8301:8301/udp \
    -p 8302:8302 \
    -p 8302:8302/udp \
    -p 8400:8400 \
    -p 8500:8500 \
    -p 53:53/udp \
    progrium/consul -server -advertise -join

Double check all 3 are present in cluster by doing curl http://localhost:8500/v1/status/peers.


Ensure every instance has:

  • A README covering available options, startup scripts, etc
  • Auto-mount point support where appropriate (eg, for saves, demos)
  • Config options for at least:
    • The name of the server (what it shows as in server browsers)
    • Server password
    • RCON password (where RCON exists)
    • LAN or internet
  • Has dockerfile configured with default ports it exposes (not that the suggested configuration uses these)
  • Has a way to easily RCON or remote control
  • Does not require user to manually download everything (with exceptions for some large and customisable downloads)
  • Fix the thing where the auto mounts always have permission problems because conatiners

Make the easy start script support passing environment variables to the container

Make an easy way to import config files in to the servers (eg mounting cfg directories)

Add more games


This project is licensed under GPL 3.0. See LICENSE for more information.