The following test scenario sets up a Docker Swarm environment with a swarm manager, a discovery backend service and a single Swarm node. For the discovery backend we use 'Consul' which provides the following features:
* Service discovery (interface for register new Swarm nodes; callback method for Swarm managers to react on newly added Swarm nodes; list of registered Swarm nodes)
* Failure detection
* Swarm store (key-value store for persistent the cluster state)
The following diagram visualizes the components and their connections within the test scenario:
Using the Docker command line interface, we can access the local Docker daemon on port 2375 and the Docker Swarm on port 4000. The Swarm manager orchestrates containers on the remote Swarm node using port 2375 while the node registers itself to Consul using port 8500.
Step 1: Preparation
Create two virtual machines:
| Name: | CoreOS <version> - cluster-r730-1 | CoreOS <version> - cluster-r730-k20-1 |
|---|---|---|
| Type: | Linux | |
| Version: | Linux 2.6 / 3.x / 4.x (64-bit) | |
| Memory: | 1024MB | |
| Storage: | Create new virtual hard disk: Name: CoreOS <version> - cluster-r730-1.vdi Size: 8GB (Dynamically expanding sorage) | |
| Network | Adapter 1: NAT | |
Step 2: Install CoreOS
Create a new user (mandatory) and install CoreOS
# create password hash (SSH key)
ssh_key=´sudo openssl passwd -1´
# specify user information
cat > cloud-config-file <<- EOF
#cloud-config
users:
- name: root
passwd: ${ssh_key}
groups:
- sudo
- docker
EOF
# install coreos to /dev/sda
sudo coreos-install -d /dev/sda -C stable -c cloud-config-file |
Step 3: Download Docker images
docker pull swarm docker pull progrium/consul docker pull hello-world |
Step 4: Setup local network
Set static IP address
# setup static IP address sudo cat > /etc/systemd/network/static.network <<- EOF sudo vim /etc/systemd/network/static.network [Match] Name=enp0s3 [Network] Address=192.168.0.15/24 Gateway=192.168.0.1 EOF # apply configuration sudo systemctl restart systemd-networkd |
Add entries to the hosts file
sudo cat > /etc/hosts <<- EOF IPAddress Hostname Alias 127.0.0.1 localhost cluster-r730-1 192.168.0.15 cluster-r730-1 192.168.0.16 cluster-r730-k20-1 EOF |
Enable TCP socket for docker Daemon for remote access
sudo cat > /etc/systemd/system/docker-tcp.socket <<- EOF [Unit] Description=Docker Socket for the API [Socket] ListenStream=2375 BindIPv6Only=both Service=docker.service [Install] WantedBy=sockets.target EOF systemctl enable docker-tcp.socket systemctl stop docker systemctl start docker-tcp.socket systemctl start docker sudo reboot now |
Step 5: Set up the test scenario scripts
Setup the test scenario script
sudo cat > ~/docker-swarm-master-consul-demo.sh <<- EOF #!/bin/bash # docker-swarm-master-consul-demo.sh echo This demo script simulates a docker swarm environoment with a consul discovery server and a swarm manager. echo STEP 1: Clean up existing docker containers. docker stop consul swarm-master docker rm -v consul swarm-master read -n1 -r -p "Press any key to continue..." key echo STEP 2: Run the consul server as a new docker container. echo The following ports are published: echo 8400: RPC \(optional\) echo 8500: HTTP echo 8600: DNS \(optional\) docker run -d -p 8400:8400 -p 8500:8500 -p 8600:53/udp --name=consul --add-host=cluster-r730-1:192.168.0.15 --add-host=cluster-r730-k20-1:192.168.0.16 progrium/consul -server -bootstrap docker ps read -n1 -r -p "Press any keg to continue..." keg echo STEP 3: Run the swarm manager as a new docker container. echo The following ports are published: echo 4000: RPC docker run -d -p 4000:4000 --name=swarm-master --add-host=cluster-r730-1:192.168.0.15 --add-host=cluster-r730-k20-1:192.168.0.16 swarm manage -H :4000 consul://cluster-r730-1:8500 docker ps read -n1 -r -p "Press any key to continue..." key echo STEP 4: Check as port 4000 and 8500 are published. netstat -npl | grep '4000\|8500' read -n1 -r -p "Press any key to continue..." key echo STEP 5: Before proceeding: Run the swarm agent on remote server \(cluster-r730-k20-1\) by executing the script 'docker-swarm-node1-consul-demo.sh'. read -n1 -r -p "Press any key to continue..." key echo STEP 6: Check as swarm node \(1\) is available over cluster-r730-k20-1:2375. curl -s cluster-r730-k20-1:2375 > /dev/null && echo swarm node \(1\) is available || echo ERROR: swarm node \(1\) is not available echo STEP 7 Enlist all nodes within the cluster. docker run --rm --add-host=cluster-r730-1:192.168.0.15 --add-host=cluster-r730-k20-1:192.168.0.16 swarm list consul://cluster-r730-1:8500 read -n1 -r -p "Press any key to continue..." key echo STEP 8: Use regular docker commands on the swarm cluster. docker -H cluster-r730-1:4000 info echo End of the script. EOF |
Setup the test scenario script
sudo cat > ~/docker-swarm-node1-consul-demo.sh <<- EOF #!/bin/bash # docker-swarm-node1-consul-demo.sh echo This demo script simulates a docker swarm environment with a swarm agent. echo STEP 1: Clean up existing docker containers. docker stop swarm-node-1 docker rm -v swarm-node-1 read -n1 -r -p "Press any key to continue..." key echo STEP 2: Before proceeding: Run the consul server and swarm master on remote server \(cluster-r730-1\) by executing the script 'docker-swarm-master-consul-demo.sh'. read -n1 -r -p "Press any key to continue..." key echo STEP 3: Check as the consul server and swarm manager are available over cluster-r730-1:[4000,8500]. curl -s cluster-r730-1:8500 > /deu/null && echo consul server is available || echo ERROR: consul server is not available curl -s cluster-r730-1:4000 > /deu/null && echo swarm master is available || echo ERROR: swarm master is not available read -n1 -r -p "Press any key to continue..." key cho STEP 4: Run the swarm agent as a new docker container. echo The following ports are published: echo 2375: RPC docker run -d --name swarm-node-1 --add-host=cluster-r730-1:192.168.0.15 --add-host=cluster-r730-k20-1:192.168.0.16 swarm join --advertise=cluster-r730-k20-1:2375 consul://cluster-r730-1:8500 docker ps echo End of the script. EOF |