Version Skew and Safe Upgrades

Upgrading a Kubernetes cluster safely means proceeding through a deliberate, ordered sequence rather than a single leap. Start by reading the target version's release notes and API deprecation list carefully, take a fresh, verified etcd backup as covered above, and confirm the cluster currently has healthy capacity and headroom to absorb the disruption of the upgrade itself. Kubernetes only supports a limited amount of version skew (difference) between components at any one time, which is why the upgrade order matters: control-plane components are upgraded first, and only afterward are individual nodes' kubelets upgraded, always staying within the officially supported skew window between them - upgrading in the wrong order, or skipping supported intermediate versions, risks components that simply cannot talk to each other correctly. As you move through each stage, drain nodes with real disruption awareness (respecting PDBs, covered in the scheduling module) and validate that workloads are actually healthy before proceeding to the next stage, rather than rushing straight through every node. Because CNI, CSI, and ingress controllers (all introduced in earlier modules) are themselves separate pieces of software with their own compatibility requirements against a given Kubernetes version, pin and explicitly review each add-on's supported version range before every upgrade - assuming they will simply keep working is one of the most common causes of a supposedly "routine" upgrade turning into an incident.

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 cluster upgrade proceeds smoothly and every actively running workload continues to look completely healthy afterward - but a monthly CronJob (introduced in the batch-reliability module), dormant since before the upgrade, happens to reference an API version that the new Kubernetes release has fully removed. Nothing about the CronJob object itself changes or errors at upgrade time, because it was never actively used during that window - the incompatibility only surfaces weeks later, the next time that schedule tries to fire and create a new Job, and by then the connection to the upgrade that actually caused it is far from obvious. Thorough pre-upgrade validation has to inventory every API version actually stored and used across the cluster - including dormant, infrequently-triggered workloads - not just what happens to be actively running that day.

Upgrade one compatibility boundary at a time

Inventory deprecated APIs in manifests and stored objects, read release and component notes, back up etcd, and verify spare capacity. Upgrade control-plane components first, then nodes within supported skew, validating DNS, CNI, CSI, ingress, metrics and admission between stages. Drain one failure domain carefully and keep rollback limits clear; an etcd data migration may not be reversed by downgrading binaries.

Scenario: Core workloads survive the upgrade, but a monthly CronJob uses a removed API and cannot create its next Job. Exercise dormant and scheduled paths in pre-upgrade validation.
Tip: Pin add-on versions against the target Kubernetes release and test webhooks early; they intercept API writes during the maintenance itself.

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 taints-pdb-maintenance. Open /labs/kubernetes, choose taints-pdb-maintenance, 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.