Beyond single Pods: how the CRI actually runs your containers, how the scheduler picks a Node, and how to put a cluster app on the real internet.

What CrashLoopBackOff, Pending, ImagePullBackOff, and OOMKilled actually mean in kubectl get pods. And the commands that take you from symptom to cause to fix.

A Pod isn't running anywhere until something decides which Node it belongs on. Filtering, scoring, taints, and affinity. The mechanics of kube-scheduler, and why Pending is a scheduling failure, not a mystery.

How Kubernetes Deployments manage ReplicaSets to scale, self-heal, and roll out zero-downtime updates — the resource you actually create in production.

Run tasks to completion instead of forever — database migrations, scheduled backups, and batch processing with Kubernetes Jobs and CronJobs.

Liveness, readiness, and startup probes — what each one actually controls, the HTTP/TCP/exec mechanisms behind them, and the classic mistake that turns a database hiccup into a crash loop.

What a ReplicaSet actually does, why Deployments manage them instead of Pods directly, and how to debug one when Pod counts don't match what you expect.

Requests and limits explained the way they actually behave in the kernel — including why CPU limits throttle pods whose average usage graph looks perfectly fine.

How kubelet talks to containerd or CRI-O through the CRI: the interface, the dockershim history, and the full path from apply to a running container.

A working mental model of a Kubernetes cluster — the control plane that decides, the nodes that run your Pods, and the reconcile loop that ties them together.

Issuers, Certificates, and one Gateway annotation: how cert-manager turns TLS issuance into desired state — plus the debug chain for when it won't issue.

The last manual step, automated: external-dns watches your HTTPRoutes and writes DNS records — to a cloud provider, or to authoritative DNS you run yourself.

Stop writing Ingress annotations. How Gateway API's role-split design works — GatewayClass, Gateway, HTTPRoute — wired end to end with Traefik, TLS included.

Every existing cluster still routes through Ingress. Read rules, pathTypes, TLS, and annotations fluently — and plan the Gateway API migration off ingress-nginx.

What actually happens when you set type: LoadBalancer — cloud controllers, MetalLB on bare metal, pending IPs, traffic policy, and the one-per-app trap.

Install Helm, the Kubernetes package manager, and connect it to your cluster. Step-by-step guide for macOS, Linux, and Windows with troubleshooting tips.

Master daily Helm commands: list, status, get values, upgrade, rollback. Covers when to use Helm vs kubectl, plus a real-world incident rollback walkthrough.

Deploy your first Helm chart to Kubernetes. Covers CI/CD pipeline charts and vendor charts with values.yaml customization, upgrades, rollbacks, and cleanup.

See how Helm translates values.yaml into running Kubernetes resources. Understand the rendering pipeline, release tracking, and how to troubleshoot each layer.

Deep dive into Kubernetes architecture. Learn how Deployments, ReplicaSets, and Pods work together when you deploy applications to K8s clusters.