Kubernetes-2-Deployments.md

Kubernetes 101 - Deployments

Deployment (official docs)

1 - Basic deployment example

Deployments are Kubernetes workloads for permanently running applications (e.g. Websites or Services). They are build on top of ReplicaSets and enable replication and rollout strategies.

A basic deployment using the nginx image looks like this:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment
  labels:
    app: nginx
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.21.4
        ports:
        - containerPort: 80

Note: The used container exposes an internal nginx webserver / proxy.

• Create the deployment
• Inspect the created pods
• What happens if you delete one of these pods manually ?

2 - Deployment rollouts and scaling

Because deployments are based on ReplicaSets they offer more advanced rollout strategies when the underlying specification has changed. This enables an application to always have a minimum availability and enables a change history and undo capabilities.

The rollout status can be monitored with the kubectl rollout status deployment/<name> command. A deployment can be restarted with kubectl rollout restart deployment/<name>. kubectl rollout history deployment/<name> inspects the history of previous changes and kubectl rollout undo deployment/<name> allows for reverting to a previous state (use --to-revision=3 for a specific revision number).

kubectl scale deployment/<name> --replicas=5 is used to change the current replica-count of a deployment.

• Scale your deployment down (from three replicas to two)
• Update your nginx deployment from nginx:1.14.2 to the current version nginx:1.21.4.
• Watch the rollout status and wait till it finishes
• Inspect the rollout history
• Undo your previous change