Introduction > Demos > Autoscaled GKE

Autoscaled GKE

Set up

Create cluster and enable cluster’s node pool autoscaling

gcloud config set compute/zone asia-southeast1-a
gcloud container cluster create autoscaled
gcloud config set container/cluster autoscaled
gcloud container node-pools update default-pool --enable-autoscaling --min-nodes=0 --max-nodes=10

Deploy pod, expose through a LB

gcloud container clusters get-credentials autoscaled
kubectl create deployment nginx --image=nginx
kubectl expose deployment nginx --port=80 --type=LoadBalancer

Configure HPA

The HPA config reference is as below

apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
  name: nginx
spec:
  maxReplicas: 50
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: nginx
  targetCPUUtilizationPercentage: 10

# cd learn-gcp/content/solutions/container/files/
kubectl apply -f nginx-hpa-autoscaled.yaml

Run a load test to scale pods and nodes up

Deploy https://pkg.go.dev/github.com/rakyll/hey. On Red Hat yum install -y hey

kubectl get svc
hey -z 20m -c 250 http://$SVC_IP

Verify

Open a terminal to view pods

kubectl get pods
kubectl get pods -w
kubectl get pods | grep nginx | wc -l
kubectl get pods | grep nginx | grep Running | wc -l

Open another terminal to view nodes

kubectl get nodes

Verify that the number of nodes scale up fast to 8, then slowly to 10.

Turn off the load test and observe the number of pods and nodes go down. Node pool will take few minutes to scale down.

Clean up

gcloud container clusters delete autoscaled

FAQs

Why default node-pool does not scale down to 0? [Answer] Few nodes are needed to run kube-system pods such as kube-dns or pod has local storage (ref. this article)

kubectl get pods -o wide -A | grep kube-dns

This default setting can be changed to remove kube-system pods blocking node’s scale down. These kube-systems pods then can be placed on other nodes. Refer to this article for more details.

kubectl create poddisruptionbudget pdb-event --namespace=kube-system --selector k8s-app=event-exporter --max-unavailable 1 &&
kubectl create poddisruptionbudget pdb-fluentd-scaler --namespace=kube-system --selector k8s-app=fluentd-gcp-scaler --max-unavailable 1 &&
kubectl create poddisruptionbudget pdb-heapster --namespace=kube-system --selector k8s-app=heapster --max-unavailable 1 &&
kubectl create poddisruptionbudget pdb-dns --namespace=kube-system --selector k8s-app=kube-dns --max-unavailable 1 &&
kubectl create poddisruptionbudget pdb-dnsauto --namespace=kube-system --selector k8s-app=kube-dns-autoscaler --max-unavailable 1 &&
kubectl create poddisruptionbudget pdb-glbc --namespace=kube-system --selector k8s-app=glbc --max-unavailable 1

How to view autoscaler logs [Answer] On Cloud Logging > Logs Explorer

resource.type="k8s_cluster"
log_name="projects/addo-demo/logs/container.googleapis.com%2Fcluster-autoscaler-visibility"
resource.labels.cluster_name="autoscaled"