Deployments

At this point, you already know the purpose of Pods and ReplicaSets. Deployments are Kubernetes objects that provide declarative updates for Pods and ReplicaSets. You can declaratively perform operations such as the following by using them:

• Perform a rollout of a new ReplicaSet.
• Change the Pod template and perform a controlled rollout. The old ReplicaSet will be gradually scaled down, whereas the new ReplicaSet will scale up at the same rate.
• Perform a rollback to an earlier version of the Deployment.
• Scale the ReplicaSet up or down.

The relationship of Deployment to ReplicaSets and Pods can be seen in the following diagram:

Note that the issue of the accidental acquisition of Pods by ReplicaSets managed by Deployments does not exist. The reason for this is that Pods and ReplicaSets use a special, automatically generated label called pod-template-hash that guarantees the uniqueness of the selection.

Let's take a look at an example Deployment manifest in the nginx-deployment.yaml file:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment-example
spec:
  replicas: 3
  selector:
    matchLabels:
      environment: test
  template:
    metadata:
      labels:
        environment: test
    spec:
      containers:
      - name: nginx
        image: nginx:1.17
        ports:
        - containerPort: 80

As you can see, the basic structure is almost identical to ReplicaSet, but there are significant differences in how Deployment behaves when you perform a declarative update. Let's quickly demonstrate this in the playground:

1. Create manually the Deployment manifest file or download it using the wget command:

wget https://raw.githubusercontent.com/PacktPublishing/Hands-On-Kubernetes-on-Windows/master/Chapter04/03_deployment-example/nginx-deployment.yaml

2. Apply the Deployment manifest file using the following command:

kubectl apply -f nginx-deployment.yaml --record

3. Wait for the Deployment to fully roll out (you can observe the number of ready Pods in your deployment using kubectl get deployment -w).
4. Now, change the Pod Spec in the template in the YAML manifest; for example, change .spec.template.spec.containers[0].image to nginx:1.16 and apply the Deployment manifest again.
5. Immediately after that, observe how the rollout progresses using the following command:

master $ kubectl rollout status deployment nginx-deployment-example
Waiting for deployment "nginx-deployment-example" rollout to finish: 1 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment-example" rollout to finish: 2 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment-example" rollout to finish: 1 old replicas are pending termination...
deployment "nginx-deployment-example" successfully rolled out

As you can see, the declarative update to the Deployment template definition caused a smooth rollout of new Pod replicas. The old ReplicaSet was scaled down and, simultaneously, a new ReplicaSet, with a new Pod template, was created and gradually scaled up. You can now try performing the same operation with an image update for an existing bare ReplicaSet and you will see that... actually, nothing happens. This is because ReplicaSet only uses a Pod template to create new Pods. Existing Pods will not be updated or removed by such a change.

Next, let's take a look at a concept similar to Deployments: StatefulSets.