KUBERNETES: TỪ CƠ BẢN ĐẾN NÂNG CAO

Module 5: Workload Management

xdev.asia

🎯 Mục tiêu bài thực hành

• Tạo Indexed Job xử lý data song song
• Cấu hình HPA với CPU metrics
• Demo In-Place Pod resizing (K8s 1.35+)
• Cài đặt KEDA và scale đến zero

Lab 1: Indexed Job — Song song xử lý Data

kubectl create namespace lab5
Indexed Job: mỗi pod xử lý 1 partition của dataset
cat <<EOF | kubectl apply -f -
apiVersion: batch/v1
kind: Job
metadata:
name: data-processor
namespace: lab5
spec:
completions: 10           # tổng 10 partitions
parallelism: 3            # chạy song song 3 pods
completionMode: Indexed   # mỗi pod có JOB_COMPLETION_INDEX
backoffLimit: 2
template:
spec:
restartPolicy: Never
containers:
- name: processor
image: busybox:1.36
command:
- sh
- -c
- |
echo "Processing partition $JOB_COMPLETION_INDEX of 10"
echo "Data: items $((JOB_COMPLETION_INDEX * 100)) to $(((JOB_COMPLETION_INDEX + 1) * 100 - 1))"
sleep $((RANDOM % 10 + 5))
echo "Partition $JOB_COMPLETION_INDEX completed!"
resources:
requests:
cpu: "100m"
memory: "64Mi"
EOF
Theo dõi job
kubectl get jobs -n lab5 -w
kubectl get pods -n lab5 -l job-name=data-processor
Xem logs của từng pod
kubectl logs -n lab5 -l job-name=data-processor --prefix=true
Kết quả
kubectl describe job data-processor -n lab5

Lab 2: CronJob với Timezone

cat <<EOF | kubectl apply -f -
apiVersion: batch/v1
kind: CronJob
metadata:
  name: morning-report
  namespace: lab5
spec:
  schedule: "0 8 * * 1-5"           # 8am thứ 2-6
  timeZone: "Asia/Ho_Chi_Minh"      # GA K8s 1.27
  concurrencyPolicy: Forbid          # không chạy 2 jobs cùng lúc
  successfulJobsHistoryLimit: 3
  failedJobsHistoryLimit: 1
  jobTemplate:
    spec:
      template:
        spec:
          restartPolicy: OnFailure
          containers:
          - name: report
            image: busybox:1.36
            command: ['sh', '-c', 'echo "Daily report at $(date)"']
EOF
kubectl get cronjob -n lab5
Trigger thủ công để test
kubectl create job --from=cronjob/morning-report morning-report-manual -n lab5
kubectl logs -n lab5 job/morning-report-manual

Lab 3: HPA với CPU Metrics

# Cài metrics-server (nếu chưa có)
kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml
Cho kind: cần --kubelet-insecure-tls
kubectl patch deployment metrics-server -n kube-system --type='json' 

-p='[{"op":"add","path":"/spec/template/spec/containers/0/args/-","value":"--kubelet-insecure-tls"}]'
Deploy ứng dụng để test HPA
cat <<EOF | kubectl apply -f -
apiVersion: apps/v1
kind: Deployment
metadata:
name: php-apache
namespace: lab5
spec:
replicas: 1
selector:
matchLabels:
app: php-apache
template:
metadata:
labels:
app: php-apache
spec:
containers:
- name: php-apache
image: registry.k8s.io/hpa-example
ports:
- containerPort: 80
resources:
requests:
cpu: "200m"
limits:
cpu: "500m"
apiVersion: v1
kind: Service
metadata:
name: php-apache
namespace: lab5
spec:
selector:
app: php-apache
ports:

port: 80
EOF

Tạo HPA
cat <<EOF | kubectl apply -f -
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: php-apache-hpa
namespace: lab5
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: php-apache
minReplicas: 1
maxReplicas: 10
metrics:

type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 50
behavior:
scaleDown:
stabilizationWindowSeconds: 300   # đợi 5 phút trước khi scale down
EOF

Generate load
kubectl run load-gen --image=busybox:1.36 -n lab5 --restart=Never -- 

sh -c "while true; do wget -q -O- http://php-apache; done"
Quan sát scaling
kubectl get hpa php-apache-hpa -n lab5 -w
kubectl get pods -n lab5 -w
Stop load
kubectl delete pod load-gen -n lab5
HPA scale down sau 5 phút (stabilizationWindow)

Lab 4: In-Place Pod Resource Updates (K8s 1.35)

# Kiểm tra phiên bản K8s hỗ trợ In-Place resizing
kubectl version --short
Deploy pod với resizePolicy
cat <<EOF | kubectl apply -f -
apiVersion: apps/v1
kind: Deployment
metadata:
name: resizable-app
namespace: lab5
spec:
replicas: 1
selector:
matchLabels:
app: resizable-app
template:
metadata:
labels:
app: resizable-app
spec:
containers:
- name: app
image: nginx:1.27
resources:
requests:
cpu: "100m"
memory: "128Mi"
limits:
cpu: "200m"
memory: "256Mi"
resizePolicy:
- resourceName: cpu
restartPolicy: NotRequired    # thay đổi CPU không cần restart
- resourceName: memory
restartPolicy: RestartContainer  # thay đổi memory cần restart
EOF
POD=$(kubectl get pods -n lab5 -l app=resizable-app -o jsonpath='{.items[0].metadata.name}')
echo "Pod: $POD"
Xem resources hiện tại
kubectl get pod $POD -n lab5 -o jsonpath='{.spec.containers[0].resources}' | python3 -m json.tool
In-place resize CPU (không restart pod!)
kubectl patch pod $POD -n lab5 --type='json' -p='[
{"op":"replace","path":"/spec/containers/0/resources/requests/cpu","value":"250m"},
{"op":"replace","path":"/spec/containers/0/resources/limits/cpu","value":"500m"}
]'
Xem pod vẫn chạy (không restart)
kubectl get pod $POD -n lab5
kubectl describe pod $POD -n lab5 | grep -A5 "Resize Status"

Lab 5: KEDA — Scale to Zero

# Cài KEDA
helm repo add kedacore https://kedacore.github.io/charts
helm repo update
helm install keda kedacore/keda --namespace keda --create-namespace
Verify
kubectl get pods -n keda
Deploy ứng dụng sẽ scale
cat <<EOF | kubectl apply -f -
apiVersion: apps/v1
kind: Deployment
metadata:
name: http-worker
namespace: lab5
spec:
replicas: 0  # bắt đầu với 0 replicas
selector:
matchLabels:
app: http-worker
template:
metadata:
labels:
app: http-worker
spec:
containers:
- name: worker
image: nginx:1.27
resources:
requests:
cpu: "100m"
memory: "128Mi"
apiVersion: v1
kind: Service
metadata:
name: http-worker
namespace: lab5
spec:
selector:
app: http-worker
ports:

port: 80
EOF

Cài KEDA HTTP Add-on (scale HTTP workloads)
helm install http-add-on kedacore/keda-add-ons-http 

--namespace keda
Tạo HTTPScaledObject
cat <<EOF | kubectl apply -f -
apiVersion: http.keda.sh/v1alpha1
kind: HTTPScaledObject
metadata:
name: http-worker-scaler
namespace: lab5
spec:
hosts:

http-worker.lab5.svc.cluster.local
targetPendingRequests: 5   # scale up khi > 5 pending requests
scaledownPeriod: 300       # scale down sau 5 phút idle
scaleTargetRef:
deployment: http-worker
service: http-worker
port: 80
replicas:
min: 0   # scale to zero!
max: 10
EOF

Test: khi gửi requests, KEDA sẽ scale từ 0 → N
for i in $(seq 1 20); do
curl http://http-worker.lab5.svc.cluster.local &
done
kubectl get pods -n lab5 -w  # xem pods scale up

Cleanup

kubectl delete namespace lab5
helm uninstall keda -n keda

Tổng kết

• ✅ Indexed Job: xử lý data song song với JOB_COMPLETION_INDEX
• ✅ CronJob với timezone support (GA K8s 1.27)
• ✅ HPA: auto scale theo CPU, stabilization window
• ✅ In-Place Pod resizing: thay đổi CPU không restart (K8s 1.35)
• ✅ KEDA: scale to zero và scale from zero