We will provide instructions of two types of OS:

• Ubuntu Xenial 16.04 (LTS)
• CentOS 7.4

Make sure the OS version is matched before continuing. Furthermore, the software dependency and network settings should be also verified before you proceed to thecd cd next step. Check the following items to prepare the environment:

• Every node has a unique MAC address and product UUID: Some plugins use the MAC address or product UUID as a unique machine ID to identify nodes (for example, kube-dns). If they are duplicated in the cluster, kubeadm may not work while starting the plugin:

// check MAC address of your NIC
$ ifconfig -a
// check the product UUID on your host
$ sudo cat /sys/class/dmi/id/product_uuid

• Every node has a different hostname: If the hostname is duplicated, the Kubernetes system may collect logs or statuses from multiple nodes into the same one.
• Docker is installed: As mentioned previously, the Kubernetes master will run its daemon as a container, and every node in the cluster should get Docker installed. For how to perform the Docker installation, you can follow the steps on the official website: (Ubuntu: https://docs.docker.com/engine/installation/linux/docker-ce/ubuntu/, and CentOS: https://docs.docker.com/engine/installation/linux/docker-ce/centos/) Here we have Docker CE 17.06 installed on our machines; however, only Docker versions 1.11.2 to 1.13.1, and 17.03.x are verified with Kubernetes version 1.10.
• Network ports are available: The Kubernetes system services need network ports for communication. The ports in the following table should now be occupied according to the role of the node:

• The Linux command, netstat, can help to check if the port is in use or not:

// list every listening port
$ sudo netstat -tulpn | grep LISTEN

• Network tool packages are installed. ethtool and ebtables are two required utilities for kubeadm. They can be download and installed by theapt-get or yumpackage managing tools.