Deployment

Deployment is a resource that provides declarative updates for Pods and ReplicaSets.

It is primarily used for managing application versions and deployments, including rolling updates.

While a ReplicaSet is responsible for checking the status and maintaining the number of Pods, a Deployment can be seen as a higher-level object that encapsulates ReplicaSets, used for managing applications at a service level.

Pods deployed as Deployment resources all have identifiers in the format <deployment-name>-<replicaset-unique-number>-<random-hash-value>.

If the replica settings are changed, or if a Pod is deleted and redeployed, its identifier will also change.

In other words, the identifiers of Pods within a Deployment change dynamically depending on the situation.

Deploying a Deployment #

Creating via CLI

kubectl create deployment nginx --image=nginx:1.14.2 --replicas=3

The above command is structured as kubectl create deployment <deployment-name> --image=<image-name> --replicas=<number-of-replicas>.
It creates a Deployment named <deployment-name> that creates and maintains Pods with the image specified by --image, in the quantity specified by --replicas.
The --image flag is mandatory. If --replicas is not specified, it defaults to 1.

Creating with a YAML file (After writing and saving the YAML file, execute kubectl apply -f <filename.yaml> in the terminal.)

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:1.14.2
        ports:
        - containerPort: 80

The value of kind must be Deployment (case-sensitive).
The number of replicas is entered in spec.replicas.
Information about the Pods to be deployed is located under spec.template.
Container image information for the Pod is entered under spec.template.spec.containers.
To manage these deployed Pods collectively, key-value pairs are assigned via the spec.selector.matchLabels field.
The key-value pairs specified in spec.selector.matchLabels and spec.template.metadata.labels must be identical.

Changing Settings of a Deployed Deployment #

Sometimes you need to update the image of a Pod already running, deployed by a Deployment,

or adjust the number of replicas.

You can change settings as follows.

Image Update #

First, in a CLI environment, you can use the kubectl set command.

The command below updates the container image for the nginx app in nginx-deployment from nginx:1.14.2 to nginx:1.16.1.

kubectl set image deployment/nginx-deployment nginx=nginx:1.16.1

Next, there's a method to modify the container image entry by fetching the YAML information of the deployed Deployment using the kubectl edit command. The item to modify is located at

.spec.template.spec.containers[0].image.

kubectl edit deployment/nginx-deployment

If you created it directly with a YAML file, you can also open that YAML file, change the image entry in spec.template.spec.containers to nginx:1.16.1, and then use the kubectl apply command.

Scaling #

If you want to easily adjust the number of replicas in a CLI environment, you can use the kubectl scale command.

Executing the command below will immediately increase the number of replicas for nginx-deployment to 5.

kubectl scale deployment/nginx-deployment --replicas=5

However, the above command only applies changes explicitly, regardless of the content of the YAML file used for the actual deployment. Since there are also methods like kubectl edit or directly modifying and applying the YAML file used for the existing deployment, it's good to choose the method according to your purpose.