How to setup a cluster with Docker Swarm

In this article, I will walk you through the process of setting up a cluster using Docker swarm.

We have 3 local or cloud machine with Ubuntu Server 16.04 and above installed. Out of these, one server will act as a manager node and other two servers will act as worker nodes.

We have configured static IP addresses for all the nodes.

192.168.0.100 for Manager node.
192.168.0.101 for Worker node01.
192.168.0.102 for Worker node02.

Each node must also be running Docker. I have used docker version 18.05.0-ce.

Install docker

The installation process is well documented elsewhere and you can read Docker’s documentation for details. The simplest method is to run the following commands on each machine:

$ curl -sSL https://get.docker.com | sh

This will install the latest version of Docker engine.

Create Swarm Manager

Creating the swarm manager node can be done with a single command. We will use host 192.168.0.100 for the manager.

$ sudo docker swarm init --advertise-addr 192.168.0.100
Swarm initialized: current node (viwovkb0bk0kxlk98r78apopo) is now a manager.
To add a worker to this swarm, run the following command:
    docker swarm join --token SWMTKN-1-3793hvb71g0a6ubkgq8zgk9w99hlusajtmj5aqr3n2wrhzzf8z-1s38lymnir13hhso1qxt5pqru 192.168.0.100:2377
To add a manager to this swarm, run 'docker swarm join-token manager' and follow the instructions.

Remember the token from the above output. This is used to join worker nodes to the manager node.

You can also see the list of nodes in your cluster.

$ sudo docker node ls
ID                             HOSTNAME         STATUS     AVAILABILITY        MANAGER STATUS
viwovkb0bk0kxlk98r78apopo  *   ubuntu-16.04     Ready      Active              Leader

Add nodes

Manager node is now ready. Next, you need to add Worker node to the Manager node.

You can do this by running docker swarm join command on both Worker node.

$ sudo docker swarm join --token SWMTKN-1-3793hvb71g0a6ubkgq8zgk9w99hlusajtmj5aqr3n2wrhzzf8z-1s38lymnir13hhso1qxt5pqru 192.168.0.100:2377
This node joined a swarm as a worker.

On the manager node, run the following command to check the node status.

$ sudo docker node ls
ID                            HOSTNAME            STATUS              AVAILABILITY        MANAGER STATUS
viwovkb0bk0kxlk98r78apopo *   managernode         Ready               Active              Leader
yf6nb2er69pydlp6drijdfmwd     workernode1         Ready               Active
yyavdslji7ovmw5fd3v7l62g8     workernode2         Ready               Active

Deploy service

With the active nodes, we can now deploy a service. Let’s create a service for nginx web server.

$ sudo docker service create -p 80:80 --name webserver --replica 3 nginx
bky6uhg2agdxeqgc2b1a5tcsm
overall progress: 3 out of 3 tasks
1/3: running   [==================================================>]
2/3: running   [==================================================>]
3/3: running   [==================================================>]
verify: Service converged

The above command will create a service named webserver and containers will be launched from docker image nginx. The service containers are deployed across all the cluster nodes.

Now, you can list and check the status of the service.

$ sudo docker service ls
ID                  NAME                MODE                REPLICAS            IMAGE               PORTS
bky6uhg2agdx        webservice          replicated          3/3                 httpd:latest        *:80->80/tcp

Verify the Service from Manager node and see whether the new container is up.

$ sudo docker service ps webservice
ID                  NAME            IMAGE               NODE                DESIRED STATE       CURRENT STATE                  ERROR           PORTS
xsa5wb0eg2ln        webserver.1     httpd:latest        managernode         Running             Running about 10 minutes ago                                       
3sbscs7m9lnh        webserver.2     httpd:latest        workernode2         Running             Running about 10 minutes ago                                      
yao2m5wi54kz        webserver.3     httpd:latest        workernode2         Running             Running about 10 minutes ago

At the moment, our web server service is running as a single instance on Docker machines.

We can scale up the service to 5 instances running on 3 machines.

$ sudo docker service scale webserver=5

Verify the Service from Manager node and see whether a new container is started or not.

$ sudo docker service ps webserver
ID                  NAME               IMAGE               NODE                DESIRED STATE       CURRENT STATE       ERROR           PORTS
xsa5wb0eg2ln        webserver.1        httpd:latest        managernode         Running             Running about 10 minutes ago                                       
3sbscs7m9lnh        webserver.2        httpd:latest        workernode2         Running             Running about 10 minutes ago                                      
yao2m5wi54kz        webserver.3        httpd:latest        workernode2         Running             Running about 10 minutes ago                                 
dfg2mswa52sf        webserver.4        httpd:latest        workernode1         Running             Running about 15 seconds ago                                 
kah1j5hs14as        webserver.5        httpd:latest        workernode1         Running             Running about 15 seconds ago

You may also use yaml file for deploying your service and save with nginx.yml

version: '3.3'
services:
  nginx:
    image: nginx:latest
    ports:
      - "80:80"

You can deploy using this yaml.

$ sudo docker stack deploy -c nginx.yml nginx

Clustering made easy with Swarm

That’s the gist of creating a Docker swarm and creating a service on your new cluster. To learn more about what Docker swarm can do, issue the command docker swarm — help to see the other commands you can use for Docker swarm.

At 47Billion, we specialize in helping companies build awesome products. If you are looking for a an experienced team to build your complete solution, please contact us at info@47billion.com.