Recently I was looking for more flexible way to ship our code in production, docker and kubernetes are the sweetheart of the devops engineers. Docker is something that let us containerise our app in an image and then we let that image to run on production. I know that we all have some experience with Virtual Machines, it is easy to get confused about the difference that will stop you from appreciating what docker does. A virtual machine works separately on top of the hypervisor of your computer on the other hand docker creates another layer of abstraction on top of your OS. It lets you share the similarities the images that you already have on your OS, and on top of that it adds another layer that has the differences. Now we can run multiple linux image on top of one machine without costing double. So it optimises, it is intelligent and it saves us. When we are running a cluster of n nodes, the complexity exponentially, what if something broke somewhere in a docker container in a cluster of n nodes? How can we ensure that which docker container should be run on which node? How do we move that docker container to another node because that node is going to be turned off for maintenance? We need a manager, who takes care of them, don’t we? Kubernetes comes along takes the responsibility.
First of all lets setup kubernetes.
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
exclude=kube*
EOF
We would turn the selinux off as the documentation says so.
setenforce 0
sed -i 's/^SELINUX=enforcing$/SELINUX=permissive/' /etc/selinux/config
Now we would be installing docker, kubeadm, and in more abstract kubernetes.
yum install -y docker kubelet kubeadm kubectl --disableexcludes=kubernetes
We would need to ensure that these are the service that should be the first thing it should be doing when the computer turns on:
systemctl enable kubelet && systemctl start kubelet systemctl enable docker && systemctl start docker
Now that we have our kubernetes and docker running on master and slave node. We would need to change one or two things in configuration for a safe initial launching on master.
vi /var/lib/kubelet/kubeadm-flags.env
KUBELET_KUBEADM_ARGS=--cgroup-driver=systemd #--network-plugin=cni
We are initializing the node on master:
kubeadm init
for future token creation
sudo kubeadm token create --print-join-command
It should generate a token for you which you would need to copy and paste on your slave node, but before that you would be need to put configuration file in proper directory with proper permission.
mkdir -p $HOME/.kube sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config sudo chown $(id -u):$(id -g) $HOME/.kube/config
on slave server
kubeadm join 10.0.15.10:6443 --token vzau5v.vjiqyxq26lzsf28e --discovery-token-ca-cert-hash sha256:e6d046ba34ee03e7d55e1f5ac6d2de09fd6d7e6959d16782ef0778794b94c61e
if you are getting something similar:
I0706 07:18:56.609843 1084 kernel_validator.go:96] Validating kernel config [WARNING RequiredIPVSKernelModulesAvailable]: the IPVS proxier will not be used, because the following required kernel modules are not loaded: [ip_vs_sh ip_vs ip_vs_rr ip_vs_wrr] or no builtin kernel ipvs support: map[ip_vs_wrr:{} ip_vs_sh:{} nf_conntrack_ipv4:{} ip_vs:{} ip_vs_rr:{}] you can solve this problem with following methods: 1. Run 'modprobe -- ' to load missing kernel modules; 2. Provide the missing builtin kernel ipvs support Pulling images required for setting up a Kubernetes cluster
running following would help
for i in ip_vs ip_vs_rr ip_vs_wrr ip_vs_sh nf_conntrack_ipv4; do modprobe $i; done
Now if you run following in server, you would know that you have some nodes attached to your kubernetes cluster.
sudo kubectl get nodes sudo kubectl describe nodes
if you had issues dealing with nodes:
kubectl apply -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml
For this demonstration I will be adding an existing python/django application on this cluster. So first of all I would be needing to dockerize that first.
at my Dockerfile
# set the base image FROM python:3.7 # File Author / Maintainer MAINTAINER Sadaf #add project files to the usr/src/app folder #set directoty where CMD will execute WORKDIR /usr/src/app ADD app_name ./app_name COPY /app_name/requirements.txt . # Get pip to download and install requirements: --no-cache-dir RUN pip install -r requirements.txt # Expose ports EXPOSE 8000 # default command to execute WORKDIR /usr/src/app/app_name RUN chmod +x app_name/gunicorn.sh CMD ./app_name/gunicorn.sh #ENTRYPOINT ["/bin/bash", "app_name/gunicorn.sh"]
Now that we have a dockerfile ready. Lets build that image:
sudo docker build -t app_name_api_server .
Time to run that image and expose it on port 8000
sudo docker run -p 8000:8000 -i -t app_name_api_server
If you like what you see on localhost:8000! Congratulations! Your app is working on docker. Now let’s push that image on docker hub. For me I have created a private repo on docker hub. To be able to push your image on docker hub you would be needing to add tags to that image first then you can push it.
sudo docker tag app_name_api_server sadaf2605/app_name_api_server sudo docker push sadaf2605/app_name_api_server
Now that you have your image pushed on docker hub. Now we will go back to our kubernetes master. As the image that we want to pull are on private image, no wonder we would need some sort of credential to pull it.
DOCKER_REGISTRY_SERVER=docker.io DOCKER_USER=Type your dockerhub username, same as when you `docker login` DOCKER_EMAIL=Type your dockerhub email, same as when you `docker login` DOCKER_PASSWORD=Type your dockerhub pw, same as when you `docker login` kubectl create secret docker-registry myregistrykey \ --docker-server=$DOCKER_REGISTRY_SERVER \ --docker-username=$DOCKER_USER \ --docker-password=$DOCKER_PASSWORD \ --docker-email=$DOCKER_EMAIL
Now lets define and yaml file that we are going to define our kubernetes deployments, app_name.yaml
.
apiVersion: apps/v1 kind: Deployment metadata: name: app-name-api-server spec: selector: matchLabels: run: app-name-api-server replicas: 1 template: metadata: labels: run: app-name-api-server spec: containers: - name: app-name-api-server image: index.docker.io/sadaf2605/app_name_api_server:latest imagePullPolicy: Always ports: - containerPort: 8000 #hostPort: 8000 env: - name: DB_USERNAME value: "user" - name: DB_PASSWORD value: "password" - name: DB_NAME value: "dbname" - name: DB_HOST value: "1.2.2.3" imagePullSecrets: - name: myregistrykey hostNetwork: true dnsPolicy: ClusterFirstWithHostNet dnsPolicy: "None" dnsConfig: nameservers: - 8.8.8.8 imagePullSecrets: - name: myregistrykey --- apiVersion: v1 kind: ConfigMap metadata: name: kube-dns namespace: kube-system data: upstreamNameservers: | ["8.8.8.8"]
Now time to run a deployment using that configuration:
sudo kubectl apply -f app_name.yaml
Lets check if we have a deployment or not:
sudo kubectl get deployments
Lets check if any instance of our docker container is running or not.
sudo kubectl get pods
Now we would be creating a service that would let us access to these pods from outside of pods.
sudo kubectl expose deployment app-name-api-server --type=LoadBalancer --name=app-name-api-server
sudo kubeadm upgrade plan --feature-gates CoreDNS=true