Scaling Pods in Kubernetes

Introduction

We're trying to persuade the dev team to use Kubernetes instead of Docker Swarm. We've got to put together a deployment in order to sell the idea to the rest of our team.

We'll have to set up a cluster with three replicas that are all running the httpd image. Once we've got that running, we're going to scale up to five replicas, and then down to two, to demonstrate how well Kubernetes pods can scale.

We're going to want three vm instance google cloud open, then in each one we'll log into the different servers using google cloud: the Kubernetes Master, and each of the existing nodes. Become root, with a sudo su -, on each of them, and we'll be ready to go.

Connect to Virtual Machines

Open three Instant Terminals on three new tabs.
Connect to each of the three Kubernete's VM's using SSH:
```
ssh cloud_user@<IP_ADDRESS>
```
Elevate your privileges to root. You will be prompted to enter the same password again:
```
sudo su -
```

Initialize the Cluster

Return to the terminal for Node 1, your Kubernetes master node

We need to initialize the cluster:

kubeadm init --pod-network-cidr=10.244.0.0/16

In the output of that command are other commands that need to run before things will work. Right off, we'll run these ones locally:

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

But the last one is something we need to run on any node that wasn't to join this cluster. It will look something like this:
```
sudo kubeadmin join <IP ADDRESS>:6443 --token <TOKEN> --discovery-token-ca-cert-hash sha256:<HASHED TOKEN>
```
Get into each of the worker nodes and paste the command in, as it appeared in the kubeadmin init command's output.

Install Flannel

Install Flannel on the master:

kubectl apply -f https://github.com/flannel-io/flannel/releases/latest/download/kube-flannel.yml

Now if we run kubectl get nodes, we should see our nodes coming online. And if we run kubectl get pods --all-namespaces, we'll see that (for the most part) all of our pods are up and running.

Create the Deployment

Let's edit our deployment configuration file (vi deployment.yml) and insert the following:

apiVersion: apps/v1
kind: Deployment
metadata:
name: httpd-deployment
labels:
    app: httpd
spec:
replicas: 3
selector:
    matchLabels:
    app: httpd
template:
    metadata:
    labels:
        app: httpd
    spec:
    containers:
    - name: httpd
        image: httpd:latest
        ports:
        - containerPort: 80

Now let's spin up the deployment:
```
kubectl create -f deployment.yml
```
We can check with a kubectl get pods

Create the Service

Next, we'll create a new service file (vi service.yml) and add the following:

kind: Service
apiVersion: v1
metadata:
name: service-deployment
spec:
selector:
    app: httpd
ports:
- protocol: TCP
    port: 80
    targetPort: 80
type: NodePort

Create the service with kubectl create -f service.yml
We can check that everything is running with kubectl get services. It looks good in the terminal, but is our web server running?
Grab the public IP from the lab page, and port from the kubectl get services command output, and try going there in a browser (http://<PUBLIC IP>:<PORT>). We should see an "It works" page.

Scale the Deployment Up and Down

In order to scale the number of replicas up or down, we need only edit deployment.yml. Change the replicas: line from 3 to 5, then apply the changes:
```
kubectl apply -f deployment.yml
```
Now if we check with kubectl get pods, we'll see there are five. But what if five is too many?
Well, we can edit that deployment.yml file again change the number of replicas to 2. Then run kubectl apply -f deployment.yml to apply the change. If we wait a bit and run another kubectl get pods, we'll see our number of pods is down to two.

END