Kubeadm is a new tool that is part of the Kubernetes distribution of 1.4.0. It allows you to install and set up a Kubernetes cluster. One of the most frequent criticisms of Kubernetes is that it’s difficult to install. Kubeadm makes this much easier, so I strongly suggest you give it a try.
Pre-requisites for creating a cluster:
- One or more machines running the compatible OS (ex: Ubuntu)
- 2-GB or more of RAM per machine
- 2-CPU or more for Master
Network connectivity among all machines in the cluster
- Install a single master Kubernetes cluster.
- Install a Pod network on the cluster so that your Pods can communicate.
Beginners can set up the pre-requisites in their own machine by creating virtual machines (VMs) in a virtual box, or they can also use multiple machines for creating clusters.
Install these requirements in each node:
$ sudo apt-get update $ sudo apt-get install -y docker.io
- Kubeadm, Kubelet, Kubectl
$ sudo apt-get update && sudo apt-get install -y apt-transport-https curl $ sudo -i $ curl -s https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add - $ cat <<EOF >/etc/apt/sources.list.d/kubernetes.list $ deb http://apt.kubernetes.io/ kubernetes-xenial main $ EOF $ exit $ sudo apt-get update $ sudo apt-get install -y kubelet kubeadm kubectl
The master is the machine where the control plane components run, including etcd (the cluster database) and the API server (which the kubectl CLI communicates with).
Before running kubeadm init in master node, first choose a pod network add-on and verify whether it requires any arguments to be passed for kubeadm initialization. Depending on which third-party provider you choose, you might need to set the --pod-network-cidr argument with kubeadm init <args>.
Configure the cgroup Driver used by kubelet
$ sudo sed -i "s/cgroup-driver=systemd/cgroup-driver=cgroupfs/g" /etc/systemd/system/kubelet.service.d/10-kubeadm.conf
$ sudo systemctl daemon-reload $ sudo systemctl restart kubelet
$ sudo kubeadm init --apiserver-advertise-address=<master-private-ip> --apiserver-cert-extra-sans=10.0.2.15 --pod-network-cidr 10.1.0.0/16 $ sudo mkdir -p $HOME/.kube $ sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config $ sudo chown $(id -u):$(id -g) $HOME/.kube/config $ sudo sysctl net.bridge.bridge-nf-call-iptables=1 $ sudo KUBECONFIG=/etc/kubernetes/admin.conf $ kubectl apply -f https://raw.githubusercontent.com/cloudnativelabs/kube-router/master/daemonset/kubeadm-kuberouter.yaml
After you finish running kubeadm init in master node, it provides the token, master-ip, sha and hash as follows:
$ kubeadm join --token <token> <master-ip>:<master-port> --discovery-token-ca-cert-hash sha256:<hash>
If you do not have the token, you can obtain it by running the following command on the master node:
$ kubeadm token list
By default, tokens expire after 24 hours. If you are joining a node to the cluster after the current token has expired, you can create a new token using the following command:
$ kubeadm token create
For reference, you can view this document: https://kubernetes.io/docs/setup/independent/create-cluster-kubeadm/
A worker node in Kubernetes was previously known as a minion. A node may be a VM or a physical machine, depending on the cluster. Each node has the services necessary to run pods and is managed by the master components.
Joining worker nodes:
To add nodes to your cluster, do the following for each machine:
- SSH to the machine
- Become root (e.g. sudo su -)
- Run the command that was returned by kubeadm init. For example:
$ kubeadm join — token <token> <master-ip>:<master-port> — discovery-token-ca-cert-hash sha256:<hash>
Now you are all set and can list the nodes from the master by running
$ kubectl get nodes
Please leave your valuable comments and questions below.