HPA with Datadog

This topic describes how to configure HPA to read pod scaling recommendations from Datadog.

note

The Smart Scaler agent chart, version 2.0.2 and later, supports Datadog to extract metrics from the Inference Agent. The HPA can then read the extracted metrics from Datadog.

Important

Please read this before you begin!

You must expose the Smart Scaler metrics to Datadog and enable external metrics in the Datadog agent of an application cluster. There can be the following possible scenarios based on the cluster location where the Smart Scaler agent, application with its corresponding HPA configuration, and the Datadog agent run:

• Scenario 1: When the Smart Scaler agent, application with its corresponding HPA configuration, and the Datadog agent run on the same cluster, then exposing the smart scaler metrics and enabling external metrics API in the Datadog agent must be done on the same cluster.

• Scenario 2: When the Smart Scaler agent runs on a separate cluster, application with its corresponding HPA configuration, and the Datadog agent run on a different cluster, then:

  • Exposing the smart scaler metrics must be done on the Smart Scaler agent cluster
  • Enabling external metrics API in the Datadog agent must be done on the Datadog agent cluster for HPA to consume the Smart Scaler metrics

info

Be sure that the Datadog agent and application with its corresponding HPA configuration are always on the same cluster.

Expose the Smart Scaler Metrics to Datadog

You must expose the Smart Scaler metrics to Datadog, from which the HPA will extract them.

Step 1: Create a ConfigMap for OpenMetrics

1. Create a conf.yaml file that points the openmetrics_endpoint to the inference-agent-service using the following template.

   init_config:
   
   instances:
     - openmetrics_endpoint: http://inference-agent-service.smart-scaler.svc.cluster.local:8080/metrics
       namespace: "smart-scaler"
       metrics:
         - smartscaler_hpa_num_pods: smartscaler_hpa_num_pods

2. Create the openmetrics-config ConfigMap in the datadog namespace using the following command:

   kubectl create configmap openmetrics-config -n datadog --from-file=conf.yaml

Step 2: Upgrade the Datadog Chart

1. Add the following settings to the existing configuration in the datadog-values.yaml file.

   note

   The ConfigMap that you created in step 2 is called out under agents.

   agents:
     volumeMounts:
       - name: openmetrics-config-1
         mountPath: /etc/datadog-agent/conf.d/openmetrics.d/
     volumes:
       - name: openmetrics-config-1
         configMap:
           name: openmetrics-config
           items:
             - key: conf.yaml
               path: conf.yaml
   clusterAgent:
     enabled: true
     metricsProvider:
       enabled: true
       useDatadogMetrics: true

2. Upgrade the Datadog chart with the updated datadog-values.yaml file with the following command:

   helm upgrade --install datadog -f datadog-values.yaml datadog/datadog -n datadog

Step 3: Verify the External Metrics API

After upgrading Datadog, you can verify if the external metrics API has been properly configured using the following steps:

1. List API services: Run the following command to list all API services across namespaces:

   kubectl get apiservice -A

2. Filter by external metrics API: Look for the v1beta1.external.metrics.k8s.io API service. You can use grep to filter the results:

   kubectl get apiservice -A | grep v1beta1.external.metrics.k8s.io

3. Verify the service status: Ensure that the service is listed as True, indicating that it is available and functioning properly:

   v1beta1.external.metrics.k8s.io        datadog/datadog-cluster-agent-metrics-api   True        11h

   In this output, datadog/datadog-cluster-agent-metrics-api should be listed with a status of True, which confirms that the Datadog external metrics API is operational after the upgrade.

Step 4: Verify the Metrics on the Datadog Dashboard

Verify the Smart Scaler metrics by visualizing them on the Datadog dashboard. The following figure illustrates the scaling recommendation metric, smartscaler_hpa_num_pods.

Enable the External Metrics API in the Datadog Agent of an Application Cluster

You must have the Smart Scaler metrics to Datadog before enabling the external metrics API to the Datadog agent of an application cluster.

Step 1: Modify the Datadog Agent Configuration

1. Add the following settings to the Datadog Agent values file to enable external metrics API:

   clusterAgent:
     enabled: true
     metricsProvider:
       enabled: true
       useDatadogMetrics: true

2. Upgrade the Datadog chart with the updated datadog-values.yaml file with the following command:

   helm upgrade --install datadog -f datadog-values.yaml datadog/datadog -n datadog

Step 2: Verify the External Metrics API

After upgrading Datadog, you can verify if the external metrics API has been properly configured using the following steps:

1. List API services: Run the following command to list all API services across namespaces:

   kubectl get apiservice -A

2. Filter by external metrics API: Look for the v1beta1.external.metrics.k8s.io API service. You can use grep to filter the results:

   kubectl get apiservice -A | grep v1beta1.external.metrics.k8s.io

3. Verify the service status: Ensure that the service is listed as True, indicating that it is available and functioning properly:

   v1beta1.external.metrics.k8s.io        datadog/datadog-cluster-agent-metrics-api   True        11h

   In this output, datadog/datadog-cluster-agent-metrics-api should be listed with a status of True, which confirms that the Datadog external metrics API is operational after the upgrade.

Step 3: Create an HPA Object to Read Recommendations

info

You must create an HPA object that points to your Datadog data source to collect pod scaling recommendations.

1. Create an HPA object file called smart-scaler-hpa.yaml that points to the smartscaler_hpa_num_pods metric to scale the deployments using the following template.

   apiVersion: autoscaling/v2
   kind: HorizontalPodAutoscaler
   metadata:
     name: <hpa name>
     namespace: <namespace>
   spec:
     scaleTargetRef:
       apiVersion: apps/v1
       kind: Deployment
       name: nginx
     behavior:
       scaleDown:
         policies:
           - type: Pods
             value: 4
             periodSeconds: 60
           - type: Percent
             value: 10
             periodSeconds: 60
         stabilizationWindowSeconds: 10
     minReplicas: 2
     maxReplicas: 30
     metrics:
       - type: External
         external:
           metric:
             name: smart_scaler.smartscaler_hpa_num_pods
             selector:
               matchLabels:
                 ss_deployment_name: "<deployment name>"
                 ss_namespace: "<namespace>"
                 ss_cluster_name: "<cluster name>"
           target:
             type: AverageValue
             averageValue: "1"

2. Apply the smart-scaler-hpa.yaml object file using the following command:

   kubectl apply -f smart-scaler-hpa.yaml -n <application-namespace>

The HPA will now collect pod scaling recommendations from your Datadog data source.