Kubernetes - Deployment and Service

New concepts have arrived

Deploying Pods with Deployment

Deployment allows you to create Pods and define specifications such as image, env, port, etc. for those Pods.
You can also define the number of replicas to manage the number of Pods for HA.
Let's create one.

Creating a deployment

vi nginx-deploy.yaml
 
apiVersion: apps/v1
kind: Deployment      # deployment > replicaset > pod
metadata:
  name: nginx-deployment       # deployment's name
  labels:
    app: nginx                 # labels can be attached to resources
spec:                 # define deployment spec
  replicas: 3         # create 3 replica pods
  selector:           # method for deployment to find pods to manage
    matchLabels:      # select pods with app=nginx label
      app: nginx
  template:                    # define meta, spec for pods to be created
    metadata:
      labels:
        app: nginx             # attach app=nginx label to pod
    spec:
      containers:              # create one container
      - name: nginx            # value of name field is nginx
        image: nginx:latest    # use latest nginx image
        ports:
        - containerPort: 80    # specify container port

Apply deployment to cluster

Apply the deploy to the cluster.

kubectl apply -f nginx-deploy.yaml

Check pod and deploy creation

The deploy will be created, and 3 pods will be created according to the spec defined in the deploy.

kubectl get pods -A  
kubectl get deploy -A

Making services accessible with Service

When only Pods are created, the state where they can be accessed locally from that node is complete.
To make them accessible from outside, you can bundle those Pods with Service.
In the previous chapter, we allowed 0.0.0.0/0 Inbound ports with NodePort Service for ports 30000~32767, and now we'll use that.

Creating a service

vi nginx-service.yaml
 
apiVersion: v1
kind: Service         # Service definition
metadata:                      # set metadata for the Service
  name: mynginxservice         # service name mynginxservice
spec:
  type: NodePort      # There are ClusterIP, LoadBalancer, NodePort (Ingress will be covered later)
  ports:
  - port: 8080        # pod ip:port
    targetPort: 80    # app ip:targetPort
    protocol: TCP     # 
    name: http        # 
  selector:
    app: nginx        # bundle nginx pods with service

Apply

kubectl apply -f nginx-service.yaml

Check service creation

kubectl get service -A

Actually access the nginx service

If you apply according to the yaml above, the NodePort will be set to a random number between 30000~32767.
Check the port with the service creation command, and you can access {worker public ip}:{node port} from any Worker node to call the nginx service.

I wondered if pods are created somewhere and if this would only call Pods on that Node.
This is because CoreDNS, a Kubernetes feature, automatically balances and calls Pods on each node when you call a Service.

MSA Kubernetes