In this tutorial, you will learn how to use Horizontal Pod Autoscaling (HPA) in Kubernetes. HPA allows you to automatically scale the number of pods in a deployment based on certain metrics, such as CPU utilization. This ensures that your application can handle increased traffic and demand without manual intervention.
Before you begin, ensure that you have the following:
- Kubernetes cluster up and running
- Metrics server up and running
kubectlcommand-line tool installed and configured to connect to your cluster
First, we need to deploy an application to demonstrate HPA. Create a file named php-apache.yaml with the following contents:
$ cat php-apache.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: php-apache
spec:
selector:
matchLabels:
run: php-apache
template:
metadata:
labels:
run: php-apache
spec:
containers:
- name: php-apache
#image: registry.k8s.io/hpa-example
image: harbor.my.com/gengwg/hpa-example
ports:
- containerPort: 80
resources:
limits:
cpu: 500m
requests:
cpu: 200m
---
apiVersion: v1
kind: Service
metadata:
name: php-apache
labels:
run: php-apache
spec:
ports:
- port: 80
selector:
run: php-apacheSave the file and apply the configuration to your cluster using the following command:
$ k apply -f php-apache.yaml
deployment.apps/php-apache created
service/php-apache createdThis will create a deployment with a single pod running a PHP Apache server.
To verify that the deployment is running, execute the following command:
kubectl get podsYou should see an output similar to the following:
NAME READY STATUS RESTARTS AGE
php-apache-xxxxxxxxx-xxxxx 1/1 Running 0 xxm
Example output:
$ k get -f php-apache.yaml
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/php-apache 0/1 1 0 5s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/php-apache ClusterIP fdf5:6da1:fe0d:cc1e::cd1f <none> 80/TCP 5s
$ k get po | grep php
php-apache-65dc947646-6h4rl 1/1 Running 0 4s
Make sure that the pod is in the "Running" state.
Now, let's configure HPA for the php-apache deployment. Run the following command:
$ kubectl autoscale deployment php-apache --cpu-percent=50 --min=1 --max=10
horizontalpodautoscaler.autoscaling/php-apache autoscaledThis command sets up HPA for the php-apache deployment with the following parameters:
--cpu-percent=50: The target CPU utilization percentage at which to scale the deployment.--min=1: The minimum number of pods the deployment should have.--max=10: The maximum number of pods the deployment can scale up to.
To check the status of HPA, execute the following command:
kubectl get hpaYou should see an output similar to the following:
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
php-apache Deployment/php-apache 0%/50% 1 10 1 xxs
The TARGETS column shows the current CPU utilization percentage and the target percentage for scaling.
To trigger the scaling of the deployment, we need to generate some load on the application. Run the following command to create a load generator pod:
$ kubectl run -i --tty load-generator --rm --image=harbor.my.com/gengwg/busybox:1.35.0 --restart=Never -- /bin/sh -c "while sleep 0.01; do wget -q -O- http://php-apache; done"This command creates a temporary pod that generates continuous requests to the php-apache service.
To monitor the autoscaling behavior, run the following command:
$ kubectl get hpa php-apache --watchYou will see the current state of the HPA, including the target CPU utilization, the minimum and maximum number of pods, and the current number of replicas. As the load increases, you should observe the HPA scaling up the number of pods to handle the increased demand.
$ kubectl get hpa php-apache --watch
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
php-apache Deployment/php-apache 0%/50% 1 10 1 40s
php-apache Deployment/php-apache 1%/50% 1 10 1 2m
php-apache Deployment/php-apache 250%/50% 1 10 1 2m30s
php-apache Deployment/php-apache 250%/50% 1 10 4 2m45s
php-apache Deployment/php-apache 250%/50% 1 10 5 3m
php-apache Deployment/php-apache 122%/50% 1 10 5 3m15s
php-apache Deployment/php-apache 87%/50% 1 10 5 3m30s
php-apache Deployment/php-apache 61%/50% 1 10 5 3m45s
php-apache Deployment/php-apache 59%/50% 1 10 7 4m
php-apache Deployment/php-apache 63%/50% 1 10 7 4m15s
php-apache Deployment/php-apache 58%/50% 1 10 7 4m30s
You can also observe the replica for the deplopyment increasing:
$ k get deploy php-apache
NAME READY UP-TO-DATE AVAILABLE AGE
php-apache 5/5 5 5 4m33s
$ k get po | grep php
php-apache-65dc947646-4t7cf 0/1 ContainerCreating 0 3s
php-apache-65dc947646-69qhj 1/1 Running 0 78s
php-apache-65dc947646-6h4rl 1/1 Running 0 4m43s
php-apache-65dc947646-ckmv9 1/1 Running 0 63s
php-apache-65dc947646-hh8t4 1/1 Running 0 78s
php-apache-65dc947646-p42vn 1/1 Running 0 78s
php-apache-65dc947646-x76hf 0/1 ContainerCreating 0 3s
$ k get deploy php-apache
NAME READY UP-TO-DATE AVAILABLE AGE
php-apache 7/7 7 7 5m42s
$ k get po | grep php
php-apache-65dc947646-4t7cf 1/1 Running 0 65s
php-apache-65dc947646-69qhj 1/1 Running 0 2m20s
php-apache-65dc947646-6h4rl 1/1 Running 0 5m45s
php-apache-65dc947646-ckmv9 1/1 Running 0 2m5s
php-apache-65dc947646-hh8t4 1/1 Running 0 2m20s
php-apache-65dc947646-p42vn 1/1 Running 0 2m20s
php-apache-65dc947646-x76hf 1/1 Running 0 65s
$ k tree deploy php-apache
NAMESPACE NAME READY REASON AGE
gengwg Deployment/php-apache - 5m13s
gengwg └─ReplicaSet/php-apache-65dc947646 - 5m13s
gengwg ├─Pod/php-apache-65dc947646-4t7cf True 33s
gengwg ├─Pod/php-apache-65dc947646-69qhj True 108s
gengwg ├─Pod/php-apache-65dc947646-6h4rl True 5m13s
gengwg ├─Pod/php-apache-65dc947646-ckmv9 True 93s
gengwg ├─Pod/php-apache-65dc947646-hh8t4 True 108s
gengwg ├─Pod/php-apache-65dc947646-p42vn True 108s
gengwg └─Pod/php-apache-65dc947646-x76hf True 33s
You can also check the HPA status:
$ k get hpa php-apache -o yaml
apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
annotations:
autoscaling.alpha.kubernetes.io/conditions: '[{"type":"AbleToScale","status":"True","lastTransitionTime":"2023-06-19T23:17:20Z","reason":"ReadyForNewScale","message":"recommended
size matches current size"},{"type":"ScalingActive","status":"True","lastTransitionTime":"2023-06-19T23:17:20Z","reason":"ValidMetricFound","message":"the
HPA was able to successfully calculate a replica count from cpu resource utilization
(percentage of request)"},{"type":"ScalingLimited","status":"False","lastTransitionTime":"2023-06-19T23:19:50Z","reason":"DesiredWithinRange","message":"the
desired count is within the acceptable range"}]'
autoscaling.alpha.kubernetes.io/current-metrics: '[{"type":"Resource","resource":{"name":"cpu","currentAverageUtilization":44,"currentAverageValue":"89m"}}]'
creationTimestamp: "2023-06-19T23:17:05Z"
name: php-apache
namespace: gengwg
resourceVersion: "636650786"
uid: 007be2b0-1d8a-433b-9f2f-13a5f0bab48e
spec:
maxReplicas: 10
minReplicas: 1
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: php-apache
targetCPUUtilizationPercentage: 50
status:
currentCPUUtilizationPercentage: 44
currentReplicas: 7
desiredReplicas: 7
lastScaleTime: "2023-06-19T23:20:50Z"
View the events:
$ k describe hpa php-apache
Name: php-apache
Namespace: gengwg
Labels: <none>
Annotations: <none>
CreationTimestamp: Mon, 19 Jun 2023 16:17:05 -0700
Reference: Deployment/php-apache
Metrics: ( current / target )
resource cpu on pods (as a percentage of request): 72% (144m) / 50%
Min replicas: 1
Max replicas: 10
Deployment pods: 7 current / 7 desired
Conditions:
Type Status Reason Message
---- ------ ------ -------
AbleToScale True ReadyForNewScale recommended size matches current size
ScalingActive True ValidMetricFound the HPA was able to successfully calculate a replica count from cpu resource utilization (percentage of request)
ScalingLimited False DesiredWithinRange the desired count is within the acceptable range
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal SuccessfulRescale 79s (x2 over 71m) horizontal-pod-autoscaler New size: 4; reason: cpu resource utilization (percentage of request) above target
Normal SuccessfulRescale 19s (x2 over 70m) horizontal-pod-autoscaler New size: 7; reason: cpu resource utilization (percentage of request) above target
$ k describe deploy php-apache
....
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal ScalingReplicaSet 39m (x2 over 109m) deployment-controller Scaled up replica set php-apache-65dc947646 to 4
Normal ScalingReplicaSet 38m (x2 over 108m) deployment-controller Scaled up replica set php-apache-65dc947646 to 7
Normal ScalingReplicaSet 31m deployment-controller Scaled down replica set php-apache-65dc947646 to 4
Normal ScalingReplicaSet 31m (x2 over 100m) deployment-controller Scaled down replica set php-apache-65dc947646 to 1
Stop the load. Observe the replicas going down.
$ kubectl run -i --tty load-generator --rm --image=harbor.my.com/gengwg/busybox:1.35.0 --restart=Never -- /bin/sh -c "while sleep 0.01; do wget -q -O- http://php-apache; done"
If you don't see a command prompt, try pressing enter.
OK!OK!OK!OK!^C
E0619 16:22:54.957208 3704281 v2.go:105] EOF
pod "load-generator" deleted
pod gengwg/load-generator terminated (Error)
NOTE: Autoscaling the replicas may take a few minutes.
$ kubectl get hpa php-apache --watch
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
php-apache Deployment/php-apache 0%/50% 1 10 1 40s
php-apache Deployment/php-apache 1%/50% 1 10 1 2m
php-apache Deployment/php-apache 250%/50% 1 10 1 2m30s
php-apache Deployment/php-apache 250%/50% 1 10 4 2m45s
php-apache Deployment/php-apache 250%/50% 1 10 5 3m
php-apache Deployment/php-apache 122%/50% 1 10 5 3m15s
php-apache Deployment/php-apache 87%/50% 1 10 5 3m30s
php-apache Deployment/php-apache 61%/50% 1 10 5 3m45s
php-apache Deployment/php-apache 59%/50% 1 10 7 4m
php-apache Deployment/php-apache 63%/50% 1 10 7 4m15s
php-apache Deployment/php-apache 58%/50% 1 10 7 4m30s
php-apache Deployment/php-apache 44%/50% 1 10 7 4m45s
php-apache Deployment/php-apache 46%/50% 1 10 7 5m
php-apache Deployment/php-apache 47%/50% 1 10 7 5m15s
php-apache Deployment/php-apache 50%/50% 1 10 7 5m30s
php-apache Deployment/php-apache 47%/50% 1 10 7 5m45s
php-apache Deployment/php-apache 45%/50% 1 10 7 6m
php-apache Deployment/php-apache 44%/50% 1 10 7 6m15s
php-apache Deployment/php-apache 21%/50% 1 10 7 6m30s
php-apache Deployment/php-apache 2%/50% 1 10 7 6m45s
php-apache Deployment/php-apache 0%/50% 1 10 7 7m1s
php-apache Deployment/php-apache 0%/50% 1 10 7 11m
php-apache Deployment/php-apache 0%/50% 1 10 3 11m
php-apache Deployment/php-apache 0%/50% 1 10 1 11m
You can also see the deployment replica back to 1.
$ k get deploy php-apache -w
NAME READY UP-TO-DATE AVAILABLE AGE
php-apache 7/7 7 7 10m
php-apache 7/3 7 7 12m
php-apache 7/3 7 7 12m
php-apache 3/3 3 3 12m
php-apache 3/1 3 3 12m
php-apache 3/1 3 3 12m
php-apache 1/1 1 1 12m
Congratulations! You have successfully set up Horizontal Pod Autoscaling (HPA) in Kubernetes. You have learned how to deploy an application, configure HPA, and test the autoscaling behavior. Autoscaling allows your application to dynamically adapt to varying levels of traffic, ensuring optimal performance and resource utilization.
HPA can also scale based on multiple metrics and custom metrics. You can specify additional metrics like memory utilization, requests per second, or any custom metric that you expose.
To scale on custom metrics, you will need to have a metrics server and custom metrics adapter set up in your cluster. The configuration and usage of these components go beyond the scope of this tutorial, but you can find detailed documentation and examples in the Kubernetes official documentation.