MasterLabSystemsMasterLabSystems

// kubernetes toolkit

Kubernetes Tools I Use Daily

Kubernetes by itself is a YAML factory. These are the seven tools that turn it into a productive platform — from inspecting pods to GitOps deploys to production observability.

Quick Reference

  • ›kubectl — fundamental CLI, scriptable
  • ›Lens / K9s — visual + TUI cluster ops
  • ›Helm — chart-based packaging and rollbacks
  • ›ArgoCD — pull-based GitOps delivery
  • ›Prometheus + Grafana — metrics and dashboards
  • ›Add `kubectx` + `kubens` for fast context/namespace switching

Learning Path

Recommended order

  1. 1.Beginner
  2. 2.Intermediate
  3. 3.Advanced

Prerequisites

  • •Docker fundamentals
  • •YAML comfort
  • •Basic networking (DNS, services, ports)

Skills you will learn

  • ✓Deploying and rolling back workloads
  • ✓GitOps with ArgoCD or Flux
  • ✓Observability with Prometheus, Grafana, Loki
  • ✓Cluster-scoped vs namespace-scoped operations

Estimated time

2–4 weeks to be comfortable; ongoing depth.

Architecture Overview

Architecture

Kubernetes Deployment Architecture

CLIENTINGRESSSERVICEPODSDATAHTTPSUsersIngressNGINX / ALBServiceClusterIPPodReplica 1PodReplica 2PodReplica 3PostgreSQLManagedRedisCache
Ingress routes external traffic to a Service that load-balances across replica Pods. Pods read config from ConfigMaps and persist via managed databases.

kubectl

The fundamental Kubernetes CLI.

Recommended

The official command-line tool. Master a handful of commands and you can operate any cluster.

Pros

  • +Universal
  • +Scriptable
  • +Works against any cluster

Cons

  • –Verbose for repetitive tasks

Best for: Everything. Required baseline.

Install kubectl

Lens

Kubernetes IDE with logs, shells, and metrics.

Desktop UI for operating clusters — view workloads, edit YAML, exec into pods, and watch metrics in one window.

Pros

  • +Beautiful UI
  • +Multi-cluster
  • +Built-in metrics

Cons

  • –Account required for some features

Best for: Daily cluster ops.

Download Lens

Helm

The package manager for Kubernetes.

Templated YAML, versioned releases, and a massive chart ecosystem (Bitnami, Prometheus, Ingress).

Pros

  • +Reusable charts
  • +Versioned rollbacks
  • +Massive ecosystem

Cons

  • –Template syntax has a learning curve

Best for: Installing third-party apps and standardizing internal services.

Install Helm

K9s

Terminal UI for blazing-fast cluster ops.

Keyboard-driven TUI for navigating namespaces, pods, logs, and deployments. Once you learn it, you stop typing kubectl.

Pros

  • +Lightning fast
  • +Keyboard-first
  • +Free

Cons

  • –Vim-like learning curve

Best for: Terminal-first SREs and DevOps.

Get K9s

ArgoCD

GitOps continuous delivery for Kubernetes.

Define your desired cluster state in Git; ArgoCD reconciles it. The standard for production K8s deploys.

Pros

  • +Pull-based GitOps
  • +Web UI
  • +Multi-cluster

Cons

  • –Requires Git discipline

Best for: Teams running production Kubernetes.

Try ArgoCD

Prometheus

The de facto standard for metrics.

Time-series database with a powerful query language (PromQL). Ships with the Spring Boot Actuator integration.

Pros

  • +Battle-tested
  • +Excellent ecosystem
  • +Free / OSS

Cons

  • –Long-term storage needs Thanos / Mimir

Best for: Application and cluster metrics.

Install Prometheus

Grafana

Dashboards and alerts for everything.

Visualize Prometheus, Loki, Tempo, and 100+ other data sources. Required companion to Prometheus.

Pros

  • +Rich dashboards
  • +Alerting
  • +Templating

Cons

  • –Dashboard sprawl if ungoverned

Best for: Observability across services and infrastructure.

Download Grafana

Commands I run every day

# Switch context fast
kubectl config use-context prod-eu

# Watch all pods in a namespace
kubectl get pods -n payments -w

# Tail logs across a deployment
kubectl logs -f -l app=orders --max-log-requests=10

# Exec into a pod
kubectl exec -it orders-7c9-xyz -- /bin/sh

# Helm upgrade with values diff
helm diff upgrade orders ./chart -f values.prod.yaml

# Sync an ArgoCD app
argocd app sync orders --prune

A production Kubernetes workflow

  1. Build image in CI → push to registry.
  2. CI bumps image tag in a Git config repo.
  3. ArgoCD detects the diff and rolls out to the cluster.
  4. Prometheus scrapes the new pods; Grafana shows golden signals.
  5. If error rate spikes, ArgoCD rolls back the last sync.

Common Mistakes

  • !Editing Deployments live with `kubectl edit` instead of through Git/Helm.
  • !Forgetting resource requests — the scheduler crams pods and noisy neighbors kill latency.
  • !Using `:latest` image tags so rollbacks are impossible.
  • !Treating ArgoCD as optional — drift between cluster and Git is the silent killer.

Production Tips

  • ★Default to 3 replicas + PodDisruptionBudget for zero-downtime upgrades.
  • ★Pin Helm chart versions and image digests, not floating tags.
  • ★Wire Prometheus → Alertmanager → Slack from day one — observability is not a phase 2.
  • ★Use NetworkPolicies to enforce namespace isolation; the cluster default is open.

Further Reading

// read next

Deploy Spring Boot to Kubernetes

// read next

Scaling Spring Boot on AWS EKS

// read next

GitOps with ArgoCD

Frequently Asked Questions

Do I need Helm if I use ArgoCD?

Yes — ArgoCD renders Helm charts. Helm handles packaging; ArgoCD handles delivery.

Lens vs K9s — which one?

Both. Lens for visual exploration, K9s when you live in the terminal.

Where do I start as a Kubernetes beginner?

Install kubectl + Lens, spin up a tiny DOKS cluster, deploy a Spring Boot Docker image via a Deployment + Service, then layer in Helm and Prometheus.