Kubernetes – Deployment object – simplifying concepts

Hi All,

GREETINGS FOR THE DAY 🙂 LIFE IS BEAUTIFUL 🙂

We are continuing series on Kubernetes – simplifying the concepts.

In previous few articles we discussed

1. What is Kubernetes ? – https://knowledge-junction.in/2021/09/03/kubernetes-introduction-starting-point-for-beginners-helps-to-prepare-the-interview/
2. Kubernetes – Components / Cluster architecture – simplifying concepts – detailed article – https://knowledge-junction.in/2022/04/27/kubernetes-components-cluster-architecture-simplifying-concepts-detailed-article/
3. Setting up local environment for Kubernetes – https://knowledge-junction.in/2021/09/11/kubernetes-simplifying-setting-up-local-environment-creating-kubernetes-cluster-locally-helps-in-interview-preparation-kubernetes-for-beginners/
4. Yaml file guide / reference – https://knowledge-junction.in/2021/09/04/complete-guide-to-yaml/
5. Kubernetes – Setting up kubernetes cluster for development on Windows 10 box using minikube – https://knowledge-junction.in/2022/05/06/kubernetes-setting-up-kubernetes-cluster-for-development-on-windows-10-box-using-minikube/
6. Kubernetes – POD – https://knowledge-junction.in/2022/05/05/kubernetes-pod/
7. Kubernetes – ReplicaSet object – simplifying concepts – https://knowledge-junction.in/2022/07/07/kubernetes-replicaset-object-simplifying-concepts/

Take away from this article

• What is Kubernetes Deployment object
• YAML specification of Deployment object
• kubectl CMDLETS for Deployment object
• What happens when Deployment object is created
• KuberNetes Deployment object states

What is Kubernetes – Deployment?

• In last article we discussed about PODs and ReplicaSet objects
• Each container (our application / docker image) is encapsulated in POD
• To have our containerized application in our Kubernetes cluster we use Kubernetes deployment object
• The Deployment object tells our Kubernetes cluster
  • How to create and update instance of our application
• Once we have deployment, Kubernetes control plane schedule (Scheduler component of Kubernetes Control Plane) application instances from deployment to run on nodes in cluster
• Once the application instances are created, Kubernetes Deployment Controller monitors those instances
  • If any node goes down, deployment controller replaces application instances with an instance on another node – This is a self-healing mechanism to address machine failure or maintenance
• Kubernetes Deployment object ensures required scalability and availability of our appliection

Deployment Yaml file

apiVersion: apps/v1
kind: Deployment
metadata:

  name: <deployment name>

spec:
  selector:
    matchLabels:
      app: <label defined in pod template - these must match>
  replicas: <number of replicas> #deployment to run 2 pods matching template
  template: <POD specification>

    metadata:
      labels:
        app: <Pods label - used in deployment selector>
    spec:#indicates pod runs one container
      containers:

      - name: <Container name>
        image: <container image>
        ports:
        - containerPort: 80

• Kind – Deployment
• apiVersion : apps/v1
• POD Template – Here we have spec.selector and spec.template are the required fields
• Here. spec.selector field how Deployment finds the Pods
• template field contains
  • .template.spec –
    • POD template – exactly the same schema as POD except api and kind
    • Pods run one container and having respective given image
• .spec.selector must match .spec.template.metadata.labels, or it will be rejected by the API.

Sample Deployment YAML file – to create the deployment for Registry image. This can be use case where we need to create our private registry. In next article we will discuss about Kubernetes registries

apiVersion: apps/v1
kind: Deployment
metadata:
  name: registry-deployment-knowledjejunction
  labels:
    app: registry
spec:
  replicas: 2
  selector:
    matchLabels:
      app: registry
  template:
    metadata:
      labels:
        app: registry
    spec:
      containers:
      - name: registry
        image: registry:2
        ports:
        - containerPort: 5000
          protocol: TCP

kubectl CMDLETS for Deployment

• Here I am using minikube utility in my local development environment
• Start the minikube – minikube start

minikube start

• Create new deployment using imperative command

kubectl create deployment <deployment name> --image=<image name>

Example:

kubectl create deployment KnowledgeJunction-Deployment --image=KnowledgeJunction

kubectl create deployment test-deploy --image=KnowledgeJunction --replicas=3

• We could also create Deployment object by using YAML file as

kubectl -f apply <Deployment YAML specification file>

• Get all Deployments

kubectl get deployments

• Get details of Deployment

kubectl describe deployments

Scaling a Deployment

• We could scale deployment using – scale parameter to kubectl CMDLET

kubectl scale <deployment-name> --replicas=5

• We could also autoscale the deployment – specify minimum and maximum number of PODs we want to run based on CPU utilization of existing PODs – we could do this using – autoscale parameter of kubectl CMDLET

kubectl autoscale <deployment-name> --min=15 --max=25 --cpu-percent=90

What exactly happen when we create Deployment object :

• Node will be searched to run the our application instance
• Our application will be scheduled to run
• Configure cluster to have new node to run application in case existing node fails or get down
• Each time new deployment found, deployment controller, ReplicaSets are created
• ReplicaSet creation event is monitored by ReplicaSet controller
• Next PODs objects are created, monitored by the Scheduler
• Scheduler selects the appropriate node for the POD based on scheduling algorithms

Deployment States

• Progressing Deployment
  • Deployment is in progressing state when
    • Deployment creates new ReplicaSet
    • Deployment is scaling up OR scaling down ReplicaSets
    • New PODs are ready or available
• Complete Deployment
  • Deployment is complete when
    • Any update requested to Deployment is completed
    • All replicas associated with Deployment are available
• Failed Deployment
  • Deployment may failed when
    • Insufficient quota
    • Image pull errors
    • Insufficient permissions

Thanks for reading !!! Please feel free to discuss / suggestions / share thoughts !!!

HAVE A GREAT TIME AHEAD !!! LIFE IS BEAUTIFUL 🙂