CKAD Application Deployment Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  It will restart the pods

  Why it's wrong here

  The HorizontalPodAutoscaler does not initiate pod restarts; it only reconciles the desired replica count of a workload, such as a Deployment or ReplicaSet. The HPA controller computes a target replica count based on observed metrics and then updates the `spec.replicas` field — it has no mechanism to delete or recreate individual pods. Restarting pods without scaling is the role of a Deployment rollout or a manual `kubectl rollout restart`, not an HPA action.

• ✗

  It will do nothing because HPA only scales based on memory

  Why it's wrong here

  The claim that HPA only scales based on memory is factually incorrect. The HPA built-in support includes standard Kubernetes resource metrics such as CPU and memory utilization, and you can even define custom or external metrics. In fact, CPU is the most commonly used and is the default metric if you do not specify one. Therefore, an HPA configured with a CPU target will definitely act when CPU crosses the threshold.

• ✓

  It will increase the number of replicas

  Why this is correct

  When the average CPU utilization across the current pods is above the target value, the HPA controller calculates a desired replica count using the formula `ceil(currentReplicas * (currentMetric / desiredMetric))`. Since the ratio is greater than 1, it produces a response that increases `spec.replicas` on the target workload. This is the exact behavior expected: scale-out to reduce the per-pod CPU load to the configured target. The controller will gradually raise the replica count, subject to the `maxReplicas` limit.

• ✗

  It will decrease the number of replicas

  Why it's wrong here

  Scaling down only occurs when the observed CPU utilization drops below the target value, making the ratio less than 1. In this scenario, the CPU is above the target, so the HPA would compute a ratio greater than 1 and therefore scale up, never down. Decreasing the replica count here would be the opposite direction and would worsen the CPU pressure, so the HPA correctly ignores that action. The direction of scaling is strictly determined by the metric-to-target ratio.