I'm a Linux systems engineer with about four years of experience working with RHEL, VMware, Docker, web servers, automation, and infrastructure. Kubernetes is new to me, so I've started Mumshad Mannambeth's course and am completing the hands-on labs alongside the videos.
I'd like advice on how to structure the rest of my learning. Is the course and its labs enough to establish a strong foundation? Should I repeatedly build clusters with kubeadm, or is it worth building one manually to understand the components in more depth? Should I prioritize cluster creation, or focus more on administration, troubleshooting, networking, storage, security, and workloads?
I'd also appreciate hearing what helped experienced engineers move from understanding Kubernetes concepts to becoming productive with it. What practice environments, projects, documentation, or exam resources would you recommend?
My goal is to become confident operating Kubernetes in production and eventually earn the relevant certification.
6 Answers
The course and labs are a good starting point, but they mainly give you the vocabulary and basic syntax. Build something that resembles a real application: expose it through Ingress, package it with Helm, use persistent storage and Secrets, and apply NetworkPolicies. Make sure networking and storage aren’t left until the very end, since failures there are common in real environments.
Security is worth practicing early rather than treating it as an advanced topic. Work with properly scoped RBAC, NetworkPolicies, Secrets, and pod security settings such as securityContext. These areas are often glossed over in beginner courses but matter a lot in production.
Build a cluster manually once so you understand what kubeadm is doing behind the scenes. After that, repeatedly rebuilding clusters has diminishing returns. Spend more time operating a cluster, troubleshooting failed deployments, and recovering from problems. For example, test what happens when a node goes down during a rollout, a disk fills up, or a workload can’t mount its volume.
For certification practice, use a realistic hands-on lab environment such as Killer.sh. It tends to be more difficult than the actual exam, but the workflow and pressure are useful preparation. The certificate can help with job screening, while troubleshooting experience is what makes you effective on the job.
A personal project deployed to a cloud or on-premises environment will teach more than simply completing another tutorial. Design a small system, define its requirements, deploy it, and keep improving it. Use kind, Minikube, k3s, or similar lightweight options if you don’t have much compute available, but try to gain experience with a more production-like cluster as well.
Troubleshooting usually teaches more than rebuilding the same cluster repeatedly. Once you understand the core objects and control-plane components, deliberately create problems and learn to investigate events, logs, scheduling issues, networking failures, and storage errors. Keep the official Kubernetes documentation close by and use it to dig deeper into anything you don’t fully understand.
That makes sense. I’ll use the manual setup to understand the architecture, then focus on keeping workloads running and diagnosing failures instead of constantly starting over.