Docker Swarm

Caution

This is an experimental feature.

Setup in Coolify

If you would like to use a Swarm to deploy your resources, you need to add the Swarm Manager to Coolify.

Optionally, you can add the Swarm Workers to Coolify. This will allow Coolify to do cleanups and other stuff on the Swarm Workers.

Docker Registry

You need to have an external Docker Registry available to use a Swarm, as all workers need to be able to pull the images you built with Coolify.

• The Swarm Manager needs to push the image to the Docker Registry.
• The Swarm Workers need to pull the image from the Docker Registry.

So set your docker login credentials accordingly. More information here.

Install Swarm Cluster

WIP This is just a brief guide to install a simple Docker Swarm cluster. For more information, please refer to the official documentation.

Prerequisites

• I will use Hetzner (referral link) for this guide. You can use any other provider.
• You need at least 3 servers to create a Docker Swarm cluster with the same architecture (ARM or AMD64).
• 1 server for the manager node.
• 2 servers for the worker nodes (you can add more worker nodes if you want).
• Add private networking to all servers if possible.

Install Docker

Install Docker on all servers. You can follow the official documentation or:

1. Install with Rancher script

curl https://releases.rancher.com/install-docker/24.0.sh | sh

2. Install with Docker script

curl https://get.docker.com | sh -s -- --version 24.0

You only need to use one of the above commands.

Configure Docker

On all servers, run the following command to start Docker.

systemctl start docker
systemctl enable docker

Caution

Hetzner specific configuration. Hetnzer servers use a MTU of 1450. You need to configure Docker to use the same MTU.

On the manager, run the following command to configure Docker.

mkdir -p /etc/docker
cat <<EOF > /etc/docker/daemon.json
{
  "default-network-opts": {
    "overlay": {
      "com.docker.network.driver.mtu": "1450"
    }
  }
}
EOF
systemctl restart docker

Create a Swarm cluster

On the manager node, run the following command to create a new cluster.

# MANAGER_IP = IP of the manager node. If you have private networking, use the private IP, like 10.0.0.x.
docker swarm init --advertise-addr <MANAGER_IP>

This command will output a command to join the cluster on the worker nodes.

It should look like something like this:

# DO NOT RUN THIS COMMAND, IT IS JUST AN EXAMPLE, HELLO!
docker swarm join --token SWMTKN-1-24zvxeydjarchy7z68mdawichvf684qvf8zalx3rmwfgi6pzm3-4ftqn9n8v98kx3phfqjimtkzx 10.0.0.2:2377

Verify the cluster

Run the following command on the manager node to verify the cluster.

docker node ls

You should see something like this:

ID                            HOSTNAME        STATUS    AVAILABILITY   MANAGER STATUS   ENGINE VERSION
ua38ijktbid70em257ymxufif *   swarm-manager   Ready     Active         Leader           24.0.2
7rss9rvaqpe9fddt5ol1xucmu     swarm-worker    Ready     Active                          24.0.2
12239rvaqp43gddtgfsdxucm2     swarm-worker    Ready     Active                          24.0.2

Deploy with persistent storage

To be able to deploy a service with persistent storage, you need to have a shared volume on the swarm-workers. So the Swarm service could move the resources between the swarm-workers.

You can always use services like AWS EFS, NFS, GlusterFS, etc.

WIP