From Containers to Orchestration

If you have used Docker, you already know the pattern: docker run starts one container on one machine, and if that machine goes down, your container goes down with it. Real applications rarely run as one container on one machine - they run as many copies (replicas) spread across many machines, called nodes, so that losing one node does not take the whole application offline. Someone, or something, has to decide which node runs which copy, notice within seconds when a copy dies, and start a replacement without a human being paged. Doing that by hand does not scale past a handful of servers, and it does not scale to 3 AM. Kubernetes is software whose entire job is that coordination: placing work onto nodes, replacing failed copies, rolling out new versions gradually, and giving applications stable network identities even as the underlying Pods (Kubernetes' term for a running instance, covered in the next module) come and go. The central idea that makes this possible is desired state. Instead of issuing a sequence of one-time commands, you write down what you want to exist - for example, "I want 3 copies of this application running" - as a small text file called an object. Kubernetes then runs background programs called controllers whose only job is to repeatedly compare what you declared against what is actually running right now, and take whatever action closes the gap. This compare-and-correct cycle is called a reconciliation loop, and it never stops running - it is not a one-time setup script, it is a permanent background process. That loop is the single biggest mental shift coming from plain Docker: you stop giving orders and start declaring intent, trusting automation to keep enforcing it indefinitely.

Analogy: Kubernetes is a thermostat, not a remote control. You declare the temperature; independent controllers measure reality and keep acting. Debugging means finding which sensor, rule or actuator prevents convergence.

Read the object as evidence

kubectl get RESOURCE NAME -n NAMESPACE -o yaml
kubectl describe RESOURCE NAME -n NAMESPACE
kubectl get events -n NAMESPACE --sort-by=.metadata.creationTimestamp
kubectl explain RESOURCE.spec

Do not memorize these as a ritual. The first command exposes desired and observed state, the second connects conditions and events, the third supplies a timeline, and the fourth checks the server's schema. Compare what the controller was asked to do with what it reports doing. Then test the narrowest hypothesis.

Scenario: A team manually restarts containers after a VM failure at 2 AM. Recovery depends entirely on which on-call engineer notices first, whether they remember the correct placement and configuration, and whether a rushed redeploy accidentally uses an outdated image tag under pressure. Kubernetes turns that fragile, error-prone human reaction into a repeatable, automatic reconciliation loop: a controller detects the missing replica within seconds and creates an exact replacement from the same declared specification, every single time, with nobody awake to do it.

Follow a reconciliation loop

Create a Deployment with kubectl create deployment web --image=nginx:1.27 --dry-run=client -o yaml, apply it, and watch kubectl get deploy,rs,pod -w. Deleting a Pod demonstrates replacement; changing the template demonstrates a rollout. The important observation is that Kubernetes does not execute a one-time script. The Deployment controller creates a ReplicaSet, the ReplicaSet controller maintains Pods, the scheduler chooses nodes, and kubelets start containers. If replacement stalls, locate the first controller whose expected child is absent instead of restarting components blindly.

Scenario: A node disappears during a release. Existing replicas elsewhere continue serving while the Deployment and scheduler collaborate to recover capacity. If all replicas shared that node, the declared replica count was present but the placement design was not resilient.

Production reasoning

Ask four questions: Who owns this object? What dependency must become ready next? Which controller reports the blocking condition? What evidence would disprove my current theory? This prevents symptom-driven changes. Record the context, namespace, object generation, image digest and recent rollout before mutation; a recreated Pod may erase the evidence you needed.

Warning: Running is not the same as ready, healthy, durable or correct. Kubernetes status is layered. Confirm the application-level outcome as well as the object state.
Goal: Put this model into practice in the Kubernetes lab namespace-basics. Open /labs/kubernetes, choose namespace-basics, predict the failure path before changing anything, then use the simulator's check command to validate the finished state.

Deliberate practice

Before the lab, write the expected object relationship and the first three commands you will run. Afterward, explain why the fix converged and name one tempting change that would only mask the symptom. Repeat using an explicit namespace and a structured output format. This prediction-observation-explanation loop is what turns command familiarity into production judgment.

45-minute investigation

  1. Map (5 min): draw the owner-to-child chain and mark every namespace, selector, identity and dependency involved.
  2. Predict (5 min): write one expected status condition, one likely event and one log or metric signal before opening the lab.
  3. Observe (10 min): collect YAML, describe output and ordered events. Do not mutate state. Record which observation disproves your first theory.
  4. Repair (15 min): make the smallest declarative correction, watch the responsible controller converge, and verify the user-facing path rather than stopping at Running.
  5. Stress (10 min): change one relevant constraint - replica count, label, readiness, resource value or placement rule - predict the outcome, observe it, then restore the known-good declaration.
Tip: Keep a short incident note with symptom, evidence, hypothesis, change, result. Across four sections this produces a reusable runbook instead of a pile of remembered commands.