luc.run

Exercise

Create the specification of a deployment named nginx with 5 replicas of pods based on the nginx:1.16 image
Create the deployment and wait for the pods to be up and running
Change the specification so the image is nginx:1.18. Update the resource and observe how the pods are updated using kubectl get pod -w
Change the deployment strategy to Recreate and update the resource
Change the specification so the image is nginx:1.20. Update the resource and observe how the pods are updated.
Delete the deployment

Documentation

https://kubernetes.io/docs/concepts/workloads/controllers/deployment/

Solution

Create the specification of a deployment with 5 replicas of pods based on the nginx:1.16 image

k create deployment nginx --image=nginx:1.16 --replicas=5 --dry-run=client -o yaml > deploy.yaml

Create the deployment and wait for the pods to be up and running

Creation of the deployment:

k apply -f deploy.yaml

Waiting for the pods to be up and running:

k get po -l app=nginx -w

Change the specification so the image is nginx:1.18. Update the resource and observe how the pods are updated.

Update the specification:

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx
  name: nginx
spec:
  replicas: 5
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - image: nginx:1.18
        name: nginx

Apply the changes:

k apply -f deploy.yaml

Observe how the pods are replaced:

k get po -l app=nginx -w

You should notice the pods are replaced one after the other following a rolling update strategy (default strategy)

Change the deployment strategy to Recreate and update the resource

You can find information on the deployment rollout strategy with the explain command:

k explain deploy.spec.strategy

Changing the strategy from RollingUpdate (default) to Recreate

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx
  name: nginx
spec:
  strategy:
    type: Recreate
  replicas: 5
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - image: nginx:1.18
        name: nginx

Apply the changes (this will not trigger the replacement of pods as the pod specification was not modified)

k apply -f deploy.yaml

Change the specification so the image is nginx:1.20. Update the resource and observe how the pods are updated.

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: nginx
  name: nginx
spec:
  strategy:
    type: Recreate
  replicas: 5
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - image: nginx:1.20
        name: nginx

Apply the changes:

k apply -f deploy.yaml

Observe how the pods are replaced:

k get po -l app=nginx -w
NAME                     READY   STATUS    RESTARTS   AGE
nginx-6d777db949-9fmnt   1/1     Running   0          10s
nginx-6d777db949-9xdwz   1/1     Running   0          10s
nginx-6d777db949-gnfsg   1/1     Running   0          10s
nginx-6d777db949-qs6k8   1/1     Running   0          10s
nginx-6d777db949-s9zgx   1/1     Running   0          10s

You should notice all the pods are terminated at the same time and new ones are created. This is the behavior of the Recreate strategy.

Delete the deployment

k delete deploy nginx