An Additive Policy Model

By default, Kubernetes networking (as described in the networking module) is wide open: any Pod can talk to any other Pod, with nothing restricting it. A NetworkPolicy object is how you actually restrict that - it selects a set of Pods using a label selector, and then declares what ingress (incoming) traffic, egress (outgoing) traffic, or both, is allowed to and from those Pods. The rule that trips people up most is this: the moment any NetworkPolicy selects a Pod for a given direction (say, ingress), that Pod becomes isolated for that direction, and any traffic not explicitly allowed by some applicable policy is now denied by default - there is no need for a separate "deny" rule, isolation is an automatic side effect of being selected at all. Because every policy only ever adds allowed traffic on top of others, and never subtracts or overrides, the order in which multiple policies are written or applied genuinely does not matter - the effective result is always simply the union of everything every applicable policy allows. One thing worth knowing before you rely on any of this: NetworkPolicy is a specification, and actually enforcing it in the real dataplane requires a CNI plugin (introduced in the networking module) that specifically implements NetworkPolicy support - not every CNI does.

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 applies a default-deny NetworkPolicy object to a namespace, expecting it to lock down traffic as intended, on a cluster whose CNI plugin does not actually implement NetworkPolicy enforcement at all. The object is accepted by the API server without any error or warning - kubectl get networkpolicy shows it exists perfectly fine - but zero traffic is actually blocked in reality, because the API accepting an object's YAML is not the same thing as some component actually enforcing its intent. Always run explicit connectivity tests against the real dataplane after applying a policy, rather than trusting that a clean kubectl apply proves anything was actually enforced.

Evaluate policy from both endpoints

A Pod becomes isolated for ingress or egress when any policy selects it for that direction. Allowed traffic is the union of applicable rules, and a connection requires source egress and destination ingress to allow it when both sides are isolated. Confirm the CNI implements the feature, then list every policy selecting each Pod and inspect actual labels.

Scenario: A new allow policy seems ineffective because another team expects first-match ordering. Policies do not override denials or run top to bottom; they contribute allowed peers and ports to a union.
Warning: A policy object accepted by the API can still be unenforced by the dataplane. Run explicit positive and negative connectivity tests as part of cluster qualification.

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 networkpolicy-kong-routing. Open /labs/kubernetes, choose networkpolicy-kong-routing, 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.