Kubernetes

Kubernetes is an open source container orchestration tool designed for automating the deployment, scaling, and handling of containerized applications. If offers high availability (no downtime), scalability (high performance) and disaster recovery (backup and restore).

Hierarchy of Kubernetes system

A Kubernetes system is composed by a cluster, which contains one or more nodes. A node is a physical server or virtual machine which can allocate one or more pods. Each pod can manage one or more containers, but normally it is only one. Note that Kubernetes does not manage containers. The smallest element Kubernetes manages is a pod.

Cluster -> Node -> Pod -> Container

Main Kubernetes components

• Pod:
• Service: It sits in front of a pod and talks to another services. It has to main functions: hold a permanent IP address and act as load balancer. The nodes communicates between them via services
• Ingress:
• Volumes:
• ConfigMap:
• Secrets:
• Deployment:
• StatefulSet:

Kubernetes architecture

A simple Kubernetes cluster is composed of one master node and one or several workers nodes.

A worker node is composed of three processes:

• Container runtime: Like Docker.
• Kubelet: Executer of ther orders received by the master scheduler.
• Kubeproxy:

A master node is responsible of: schedule, re-schedule/re-restart, monitor pods and join new nodes. A master node is composed of four processes:

• API Server:
• Scheduler:
• Controller Manager:
• etcd: is the cluster brain. It contains the status of the whole cluster. Cluster changes are stored in the key value store.

Deployment and layers of abstraction

Deployment manages a ReplicaSet. A ReplicaSet manages a Pod. A Pod is an abstraction of a Container.

Deployment is the way to launch Pods. Everthing below Deployment is handled automatically by Kubernetes. This means that once you launch a Deployment there is no further configuration you can do to ReplicaSets and Pods directly.

• CRUD commands:

kubectl create deployment [NAME]
kubectl edit deployment [NAME]
kubectl delete deployment [NAME]

• Status of different k8s components:

kubectl get nodes | pod | services | replicaset | deployment

• Debugging pods:

kubectl logs [pod name]
kubectl exec -it [pod name] --/bin/bash
kubectl describe pod [pod name]

• Configuration file for CRUD:

kubectl apply -f [filename]
kubectl delete -f [filename]

Configuration file

Each configuration file has 3 parts:

• Metadata
• Specification
• Status

Each configuration file can be written for one of the k8s components:

• Deployment
• Service

Tutorial

Launch three VMs with Vagrant

For this tutorial you need to install Vagrant and Virtual Box in your local machine.

We'll generate three virtual machines with Centos 7.6 installed using a Vagrantfile.

Also, in the three VMs, we will install Docker, and disable selinux and firewalld using this file: playbook_centos_install_docker.yaml

1. Get the files:

mkdir MyCluster && cd MyCluster
wget https://raw.githubusercontent.com/cheomanigua/webdevdocs/master/aux/Vagrantfile
wget https://raw.githubusercontent.com/cheomanigua/webdevdocs/master/aux/playbook_centos_install_docker.yaml

2. Create the VMs:

vagrant up

3. How to access the VMs:

vagrant status
vagrant ssh master
or
vagrant ssh worker1
or
vagrant ssh worker2

4. Login as root user once inside the VM:

su -

The password is vagrant for all the three VMs.

Common configuration

The next steps have to be performed logged as root in all the three VMs

1. Configure Kubernetes repository

Copy the whole text and paste it in the command line:

cat <<EOF > /etc/yum.repos.d/kubernetes.repo
[kubernetes]
name=Kubernetes
baseurl=https://packages.cloud.google.com/yum/repos/kubernetes-el7-x86_64
enabled=1
gpgcheck=1
repo_gpgcheck=1
gpgkey=https://packages.cloud.google.com/yum/doc/yum-key.gpg https://packages.cloud.google.com/yum/doc/rpm-package-key.gpg
EOF

2. Install kubelet, kubeadm and kubectl

yum install -y kubelet kubeadm kubectl
systemctl enable kubelet
systemctl start kubelet

3. Disable SWAP

sed -i '/swap/d' /etc/fstab
swapoff -a

4. Set the net.bridge.bridge-nf-call-iptables to ‘1’ in your sysctl config file. This ensures that packets are properly processed by IP tables during filtering and port forwarding.

echo 'net.bridge.bridge-nf-call-ip6tables = 1' >> /etc/sysctl.conf
echo 'net.bridge.bridge-nf-call-iptables = 1' >> /etc/sysctl.conf
sysctl -p

How to Deploy a Kubernetes Cluster

The next steps have to be performed logged as root in master VM

1. Initiate kubeadm

kubeadm init --ignore-preflight-errors=SystemVerification --apiserver-advertise-address=192.168.2.10 --pod-network-cidr=10.244.0.0/16

Don't forget to copy and keep the 'kubeadm join' token and hash

2. Configure kubectl

mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

3. Configure flannel

cd /root
curl -Lo kube-flannel.yml https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml
sed -i.bak -e "s/ip-masq/ip-masq\\n - --iface=eth1/g" kube-flannel.yml

IMPORTANT: Edit kube-flannel.yml and find all the lines with '- --iface=eth1'. These lines have tab indentation, which will yield an error when running kubectl create. Remove the tab indentation and add space indentation instead.

Finally:

kubectl create -f kube-flannel.yml

It will take a minute or so for the pods and nodes to be ready.

4. Check status of cluster

kubectl get pods --all-namespaces
kubectl get nodes

Adding a Worker Node to Cluster

1. Login as root in worker1 node

vagrant ssh worker1
su -

The password is vagrant

2. Join worker1 node to cluster

kubeadm join 192.168.2.10:6443 --token blablabla --discovery-token-ca-cert-hash sha256:blablabla

The token and hash are the ones generated when running kubeadm init

You can repeat the steps for the worker2 node.

Once finished, you can check the status of the cluster from the master node:

su -
kubectl get pods --all-namespaces
kubectl get nodes

Removing a Worker Node from the Cluster

kubectl drain worker2 --delete-local-data --ignoredaemonsets

Delete Worker Node definitely from Cluster

kubectl delete node worker2

Generate new token

If the token has expired, you can generate a new one:

kubeadm token create --print-join-command

You can list all tokens:

kubeadm token list

Deploying an application to the Cluster

From the master node as root, issue:

kubectl run my-nginx --image=nbrown/nginxhello:1.12.1 --port=80
kubectl get pod
kubectl expose pod my-nginx --type=NodePort --name my-nginx-service
kubectl get service
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kubernetes ClusterIP 10.96.0.1 <none> 443/TCP 21m
my-nginx-service NodePort 10.109.2.145 <none> 80:31606/TCP 32s

We can visit the application at http://192.168.2.11:31606