Introduction to Kubernetes: Deploy Your First App in Just 5 Minutes of Tinkering – ITFROMZERO

Back in 2018, I used to stay up until 4 AM just to SSH into over 20 servers, manually pulling new code and restarting services. At the time, I thought I was an operations ‘superhero’—until a single node failure triggered a cascading system crash. That was the moment I realized: I needed immediate automation. And Kubernetes (K8s) was the answer.

K8s might sound intimidating, but imagine it as a high-level ‘orchestrator‘ for your infrastructure. You simply give the command: “I want exactly 5 replicas of this web app running.” The rest—from selecting available servers to self-healing after a failure—this orchestrator handles from A to Z.

Table of Contents

Hands-on: Running Your First App in 5 Minutes

Instead of cramming theory, let’s dive straight into practice. Install Minikube (a lightweight K8s for your laptop) or use Play with Kubernetes to practice online for free.

Once you have the kubectl command-line tool ready, try this ‘magical’ sequence of commands:

# 1. Create a Deployment running the Nginx image
kubectl create deployment hello-k8s --image=nginx

# 2. Check if the Pod is 'up and running'
kubectl get pods

# 3. Expose the port for browser access
kubectl expose deployment hello-k8s --type=NodePort --port=80

# 4. Get the URL to see the result (Minikube only)
minikube service hello-k8s --url

In less than 120 seconds, you’ve deployed a container to the cluster. K8s automatically finds the optimal spot on your infrastructure to place Nginx, ensuring it remains in a ready state.

Breaking Down the Concepts: Pod, Deployment, and Service

If those terms sound confusing, let me explain them in the most ‘down-to-earth’ way possible.

Pod – The Smallest Unit

In the Docker world, we talk about Containers. But in K8s, the smallest unit is a Pod. Think of a Pod like an apartment, where containers are roommates sharing utilities (resources). Gold rule: Each Pod should typically only contain one primary container for the most effective management.

Deployment – The Dedicated Manager

Never create Pods manually. Use a Deployment instead. This is where you declare your ‘Desired State.’ For example: “There must always be 3 stable Pods running.” If a Pod suddenly ‘dies,’ the Deployment immediately initializes a new Pod on another server to compensate. The system self-heals without you needing to stay up late and press buttons.

Service – The Permanent Hotline

Pods in K8s are ‘fragile.’ They can be deleted and recreated with constantly changing IP addresses. So, how do users find your app? The Service is the answer. It provides a fixed IP address and domain name, acting as a shield to accurately route traffic to healthy Pods.

Levelling Up: Scaling Applications with YAML Files

The true power of K8s lies in its declarative nature. Instead of typing commands, you write everything in a YAML file. Here’s how I scale the system to 3 replicas with extreme ease:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment
spec:
  replicas: 3 # Scale up instantly here
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.21
        resources:
          limits:
            memory: "256Mi"
            cpu: "500m"

Simply run kubectl apply -f deployment.yaml, and K8s will automatically ‘compare’ this file with the current state. It then performs the necessary changes to bring the system in line with your requested configuration.

A Few Tips to Save You from Trouble

After years of working with various clusters, I’ve gathered 4 hard-learned lessons for you:

Never Leave Resources Unbounded: Don’t forget the limits section. I once had a buggy script with a memory leak crash an entire node running 15 other services simply because I forgot to cap the RAM at 256MB.
Say No to the :latest Tag: Specify a concrete version like nginx:1.21. Using latest makes it impossible to know exactly which version is running, leading to major headaches during debugging.
Be Friends with Logs: The kubectl logs -f [pod-name] command is your primary lifesaver. When the app isn’t running, go there to hear its ‘complaints’ and find out how to fix it.
Double-check your YAML: YAML files are extremely sensitive to whitespace. Install a YAML extension on VS Code immediately to avoid wasting hours over a single extra space.

K8s has many more exciting features like Ingress, ConfigMap, and Secrets waiting for you to discover. Don’t be afraid to make mistakes and start over; that’s the fastest way to master this technology. If you get stuck anywhere, just leave a comment below, and I’ll do my best to help!