Resource Requirements, Limits, and Quotas

QoS classes

Requests vs limits

• Scheduler sums all container requests in Pod
• HPA needs CPU requests: else <unknown> target

resources:
  requests:
    cpu: 250m
    memory: 64Mi
  limits:
    memory: 64Mi    # best practice: memory limit = request
    cpu: 500m

ResourceQuota (namespace aggregate)

• Caps total resources used in a namespace

apiVersion: v1
kind: ResourceQuota
metadata:
  name: compute-quota
spec:
  hard:
    requests.cpu: "4"
    requests.memory: 8Gi
    limits.cpu: "8"
    pods: "10"

kubectl create quota compute-quota --hard=requests.cpu=4,requests.memory=8Gi,pods=10
kubectl describe quota compute-quota

• Pod rejected if quota exceeded: check events

LimitRange (per-object defaults/constraints)

• Default/min/max per Pod or container in namespace

apiVersion: v1
kind: LimitRange
metadata:
  name: mem-limit-range
spec:
  limits:
  - default:
      memory: 512Mi
    defaultRequest:
      memory: 256Mi
    max:
      memory: 1Gi
    min:
      memory: 128Mi
    type: Container

In-place Pod vertical scaling

Kubernetes v1.35 includes in-place CPU/memory resizing support where the feature is enabled. For CKAD, know the idea and how to inspect it; most tasks still use Pod templates on Deployments.

kubectl set resources deployment/app \
  --requests=cpu=250m,memory=256Mi \
  --limits=cpu=500m,memory=512Mi

kubectl get pod <pod> -o jsonpath='{.status.containerStatuses[*].resources}'

• Deployment template change is the safe exam path
• Pod-level resize depends on cluster feature support and resource resize policy

Failures

Exam tips

• Pending + PodExceedsFreeCPU: requests too high for any node
• ResourceQuota vs LimitRange: namespace total vs per-Pod/container
• Without requests on containers, HPA cannot scale on CPU %
• In-place resize appearing in a question is likely inspect/patch-level, not deep scheduling theory