Kubernetes Resource Requests Limits

resources:
  requests:
    cpu: 200m       # Guaranteed minimum
    memory: 256Mi   # Guaranteed minimum
  limits:
    cpu: "1"        # Maximum allowed
    memory: 512Mi   # Maximum allowed (OOMKilled if exceeded)

cpu: "1"       # 1 vCPU core
cpu: "0.5"     # Half a core
cpu: 500m      # 500 millicores = 0.5 cores
cpu: 100m      # 100 millicores = 0.1 cores
cpu: 250m      # Quarter core

memory: 128Mi   # 128 mebibytes (128 × 1024² bytes)
memory: 256Mi
memory: 1Gi     # 1 gibibyte
memory: 512M    # 512 megabytes (decimal, not binary)

# Requests == Limits for ALL containers
resources:
  requests:
    cpu: 500m
    memory: 256Mi
  limits:
    cpu: 500m
    memory: 256Mi

# Requests < Limits (or only requests set)
resources:
  requests:
    cpu: 100m
    memory: 128Mi
  limits:
    cpu: 500m
    memory: 512Mi

# No requests or limits set
resources: {}

resources:
  requests:
    cpu: 100m
    memory: 128Mi
  limits:
    cpu: 500m
    memory: 256Mi

resources:
  requests:
    cpu: 250m
    memory: 256Mi
  limits:
    cpu: "1"
    memory: 512Mi

resources:
  requests:
    cpu: 500m
    memory: 512Mi
  limits:
    cpu: "2"
    memory: 1Gi

resources:
  requests:
    cpu: "1"
    memory: 2Gi
  limits:
    cpu: "1"
    memory: 2Gi

# Current usage
kubectl top pods
kubectl top pods --containers

# Detailed per-pod
kubectl top pod my-pod --containers

# Average CPU usage over 24h
avg_over_time(
  rate(container_cpu_usage_seconds_total{pod="my-pod"}[5m])[24h:]
)

# Peak memory usage over 24h
max_over_time(
  container_memory_working_set_bytes{pod="my-pod"}[24h]
)

# CPU request vs actual usage (over-provisioning)
sum(kube_pod_container_resource_requests{resource="cpu"})
/ sum(rate(container_cpu_usage_seconds_total[5m]))

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: api-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: api
  updatePolicy:
    updateMode: "Off"  # Start with recommendations only
  resourcePolicy:
    containerPolicies:
      - containerName: api
        minAllowed:
          cpu: 50m
          memory: 64Mi
        maxAllowed:
          cpu: "2"
          memory: 2Gi

# Check recommendations
kubectl describe vpa api-vpa

kubectl describe pod my-pod | grep -A5 "Last State"
# Reason: OOMKilled
# Exit Code: 137

# Fix: increase memory limit
kubectl set resources deployment my-app --limits=memory=1Gi

# Check throttling
kubectl exec my-pod -- cat /sys/fs/cgroup/cpu/cpu.stat
# nr_throttled: 12345  ← high number = significant throttling

# Fix: increase CPU limit (or remove it)
kubectl set resources deployment my-app --limits=cpu=2

kubectl describe pod my-pod
# Events:
# FailedScheduling: Insufficient cpu / Insufficient memory

# Check node capacity
kubectl describe nodes | grep -A5 "Allocated resources"