Trace Denied Traffic

When traffic that should be allowed appears blocked, resist immediately blaming NetworkPolicy - first confirm the fundamentals covered in the networking module actually work independently of any policy: does the destination Service have ready endpoints, and does DNS resolve correctly? Only once those check out should you move to inspecting policy specifically: list every NetworkPolicy that selects the source Pod (checking its egress rules) and every one that selects the destination Pod (checking its ingress rules) - remembering from the additive model above that a connection needs both sides to permit it if both are isolated. A reliable way to test is spinning up a disposable, ephemeral client Pod deliberately labeled to match the real source's labels exactly, since a policy written around specific labels will behave completely differently for a debug Pod that does not carry them. Change only one rule at a time as you test, inspect the destination's ingress rules and the source's egress rules separately (a failure could originate from either side), and where available, check your CNI's own logs or connection counters, which some plugins expose specifically to show whether a policy actively dropped a given connection.

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: An ingress controller like Kong is explicitly permitted, by policy, to reach the destination application's Service CIDR range - which sounds like it should cover the traffic - but NetworkPolicy rules are actually evaluated against the selected Pod's real endpoint IPs, not against the Service's virtual ClusterIP address at all, since the policy only ever sees traffic once it has already been routed to an actual Pod. Connections still fail because the rule, written against the wrong address concept, never actually matches the real destination Pods. Express destination rules using Pod and namespace selectors - matching what NetworkPolicy actually evaluates - rather than a Service's CIDR, which is not the address policy enforcement ever sees.

Reproduce the denied flow faithfully

Record source namespace, labels and ServiceAccount; destination Pod labels and port; DNS result; and whether traffic goes through a gateway or Service. Create a debug Pod with matching labels, then test name, ClusterIP and direct endpoint. Inspect both direction policies and CNI flow logs or counters. Change one rule at a time and retain a negative test.

Scenario: Direct endpoint access works but Service access fails only from one node. Policies are identical, so investigate Service dataplane state on that node rather than adding selectors.
Tip: A successful curl after allowing 0.0.0.0/0 proves only that policy caused the block. Narrow systematically to the intended peer and port before declaring the repair complete.

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.