This page describes how to set up App and API Protection for Kubernetes to automatically configure supported Kubernetes ingress proxies and gateways to run API discovery, threat detection, and inline blocking at the edge of the infrastructure.

Overview

App and API Protection for Kubernetes automatically configures supported ingress proxies and gateways in your Kubernetes cluster to enable Application Security monitoring. This eliminates the need for manual proxy configuration and provides API-wide security coverage without modifying individual services or deploying tracers across your application fleet.

What performs the automatic configuration?

App and API Protection for Kubernetes uses a Kubernetes controller (running in the Datadog Cluster Agent) that:

• Automatically detects supported proxies in your cluster
• Configures proxies to route traffic through an external Application Security processor
• Enables threat detection for all traffic passing through your ingress layer
• Simplifies operations through centralized configuration with Helm

Supported proxies

For the list of supported proxies and proxy-specific setup steps, see the setup page.

Limitations

Sidecar mode

• Requires Datadog Cluster Agent 7.80.2 or later
• Each gateway pod runs its own processor instance, which increases per-pod resource usage

External mode

• Requires Datadog Cluster Agent 7.80.2 or later
• Security processor must be manually deployed and scaled
• Deployed service may require an appropriate network policy:
  • From the proxy pods on the service port
  • To the Datadog Agent for traces

Proxy compatibility

• For proxy version compatibility, see the compatibility documentation.

Prerequisites

Before enabling App and API Protection for Kubernetes, verify that you have:

• A running Kubernetes cluster (version 1.20 or later)
• Datadog Cluster Agent 7.80.2 or later installed and configured in your cluster
• One or more supported proxies installed
• Remote Configuration enabled to allow blocking attackers through the Datadog UI

How it works

App and API Protection for Kubernetes supports two deployment modes:

• Sidecar mode (default): The Application Security processor runs as a sidecar container injected directly into each gateway pod. No separate processor deployment is needed, and the processor scales automatically with your gateway pods.
• External mode: A single, centralized Application Security processor deployment serves all gateway traffic in your cluster. Use this mode when you want to manage one shared processor for the whole cluster.

To set up the default sidecar mode, see Set up sidecar mode. To deploy a centralized processor instead, see Set up external mode.

Set up sidecar mode

In sidecar mode, the security processor runs as a container injected directly into each gateway pod. The Cluster Agent handles injection automatically, so you don’t need a separate processor deployment or service.

When to use sidecar mode

• You prefer not to manage a separate processor deployment and service
• You want the processor co-located with each gateway pod

Setup

datadog:
  appsec:
    injector:
      enabled: true
      # mode defaults to "sidecar" when omitted

helm upgrade -i datadog-agent datadog/datadog -f values.yaml

apiVersion: datadoghq.com/v2alpha1
kind: DatadogAgent
metadata:
  name: datadog
  annotations:
    agent.datadoghq.com/appsec.injector.enabled: "true"

kubectl apply -f datadog-agent.yaml

Sidecar configuration reference

All sidecar parameters are available as Helm values nested under datadog.appsec.injector.sidecar, or as DatadogAgent annotations (Datadog Operator version 1.27.1 or later):

sidecar.image

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.image

Type: String

Default: ghcr.io/datadog/dd-trace-go/service-extensions-callout

Description: Sidecar container image

sidecar.imageTag

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.image_tag

Type: String

Default: v2.6.0

Description: Sidecar container image tag

sidecar.port

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.port

Type: Integer

Default: 8080

Description: gRPC listening port for the sidecar processor

sidecar.healthPort

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.health_port

Type: Integer

Default: 8081

Description: Health check port for the sidecar processor

sidecar.bodyParsingSizeLimit

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.body_parsing_size_limit

Type: Integer

Default: 0

Description: Maximum request body size in bytes to process. 0 disables body processing. Use -1 to disable body parsing entirely.

sidecar.resources.requests.cpu

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.resources.requests.cpu

Type: String

Default: 10m

Description: CPU request for the sidecar container

sidecar.resources.requests.memory

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.resources.requests.memory

Type: String

Default: 128Mi

Description: Memory request for the sidecar container

sidecar.resources.limits.cpu

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.resources.limits.cpu

Type: String

Default: ""

Description: CPU limit for the sidecar container (optional)

sidecar.resources.limits.memory

Datadog Operator annotation: agent.datadoghq.com/appsec.sidecar.resources.limits.memory

Type: String

Default: ""

Description: Memory limit for the sidecar container (optional)

Set up external mode

In external mode, you deploy a single, centralized Application Security processor that serves all gateway traffic in your cluster. The Cluster Agent automatically configures your supported proxies to route traffic to this processor.

Architecture

• Security Processor Deployment: You deploy a centralized Application Security processor as a Kubernetes Deployment with an associated Service.
• Automatic Proxy Detection: The controller watches for supported proxy resources in your cluster using Kubernetes informers.
• Automatic Configuration: When proxies are detected, the controller creates the proxy configuration needed to route traffic to the security processor service.
• Traffic Processing: Gateways route traffic to the security processor through the Kubernetes service for security analysis.

Benefits

• Resource Efficient: A single shared processor handles traffic from all gateways
• Centralized Management: One deployment to monitor, scale, and configure
• Infrastructure-as-Code: Manage configuration through Helm values
• Non-Invasive: No application code changes required
• Scalable: Add new gateways without additional configuration

Step 1: Deploy the security processor

Deploy the security processor service, which analyzes traffic forwarded from your gateways. For proxy-specific deployment details, see the setup documentation for your proxy.

Example deployment:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: datadog-aap-extproc-deployment
  namespace: datadog
spec:
  replicas: 2
  selector:
    matchLabels:
      app: datadog-aap-extproc
  template:
    metadata:
      labels:
        app: datadog-aap-extproc
    spec:
      containers:
      - name: datadog-aap-extproc-container
        image: ghcr.io/datadog/dd-trace-go/service-extensions-callout:v2.4.0
        ports:
        - name: grpc
          containerPort: 443
        - name: health
          containerPort: 80
        env:
        # Use the address of the datadog agent service in your cluster
        - name: DD_AGENT_HOST
          value: "datadog-agent.datadog.svc.cluster.local"

        - name: DD_SERVICE_EXTENSION_TLS
          value: "false"
        readinessProbe:
          httpGet:
            path: /
            port: health
          initialDelaySeconds: 5
          periodSeconds: 10
        livenessProbe:
          httpGet:
            path: /
            port: health
          initialDelaySeconds: 15
          periodSeconds: 20
---
apiVersion: v1
kind: Service
metadata:
  name: datadog-aap-extproc-service
  namespace: datadog
spec:
  ports:
  - name: grpc
    port: 443
    targetPort: grpc
  selector:
    app: datadog-aap-extproc
  type: ClusterIP

Apply the manifest:

kubectl apply -f datadog-aap-extproc-service.yaml

Step 2: Enable automatic configuration

Point the Datadog Cluster Agent at your security processor service using Helm or the Datadog Operator.

Note: The processor service name (datadog-aap-extproc-service) must match the service you deployed in Step 1.

apiVersion: datadoghq.com/v2alpha1
kind: DatadogAgent
metadata:
  name: datadog
  annotations:
    agent.datadoghq.com/appsec.injector.enabled: "true"
    agent.datadoghq.com/appsec.injector.mode: "external"
    agent.datadoghq.com/appsec.injector.processor.service.name: "datadog-aap-extproc-service"  # Required: must match your security processor service name
    agent.datadoghq.com/appsec.injector.processor.service.namespace: "datadog"

kubectl apply -f datadog-agent.yaml

datadog:
  appsec:
    injector:
      enabled: true
      mode: "external"
      processor:
        service:
          name: datadog-aap-extproc-service  # Required: must match your security processor service name
          namespace: datadog                 # Must match the namespace where the service is deployed

helm upgrade -i datadog-agent datadog/datadog -f values.yaml

Step 3: Verify the installation

Check that the Cluster Agent detected your proxies:

kubectl logs -n datadog deployment/datadog-cluster-agent | grep appsec

Verify proxy configuration

Verify that the controller created the proxy configuration resources for your proxy. For proxy-specific verification commands, see the setup documentation for your proxy.

The Datadog Cluster Agent produces events for each operation that results in failure or success done in the cluster.

Test traffic processing

Send requests through your gateway and verify they appear in the Datadog App and API Protection UI:

1. Navigate to Security > Application Security in Datadog.
2. Look for security signals from your gateway traffic.
3. Verify that threat detection is active.

Configuration reference

Automatic configuration options

enabled

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.enabled

Type: Boolean

Default: false

Description: Enable or disable the integration

mode

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.mode

Type: String

Default: ""; when empty, defaults to sidecar

Description: Injection mode: "sidecar" or "external"

autoDetect

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.autoDetect

Type: Boolean

Default: true

Description: Automatically detect and configure supported proxies

proxies

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.proxies

Type: JSON array

Default: []

Description: Manual list of proxy types to configure. For valid values, see the setup page.

processor.service.name

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.processor.service.name

Type: String

Default: None

Description: Required. Name of the security processor Kubernetes Service

processor.service.namespace

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.processor.service.namespace

Type: String

Default: Defaults to the namespace where the Cluster Agent is running

Description: Namespace where the security processor service is deployed

processor.address

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.processor.address

Type: String

Default: {service.name}.{service.namespace}.svc

Description: Full service address override

processor.port

Datadog Operator annotation: agent.datadoghq.com/appsec.injector.processor.port

Type: Integer

Default: 443

Description: Port of the security processor service

Upgrading from external mode

If you are upgrading from a previous version that used external mode, the default mode has changed to sidecar. To continue using external mode, explicitly set mode: "external" in your Helm values:

datadog:
  appsec:
    injector:
      enabled: true
      mode: "external"
      processor:
        service:
          name: datadog-aap-extproc-service
          namespace: datadog

Opting out specific resources

You can exclude specific Gateway or GatewayClass resources from automatic configuration by adding a label:

apiVersion: gateway.networking.k8s.io/v1
kind: Gateway
metadata:
  name: my-gateway
  namespace: my-namespace
  labels:
    appsec.datadoghq.com/enabled: "false"  # Exclude this gateway from automatic configuration
spec:
  # ... gateway configuration

Resources with the appsec.datadoghq.com/enabled: "false" label are ignored. This is useful when you want to:

• Manually configure specific gateways
• Temporarily disable App and API Protection for testing
• Exclude certain gateways from security monitoring

Note: By default, all resources are included. Only resources with the label explicitly set to "false" are excluded.

Troubleshooting

All errors are logged as Kubernetes events. Check for events on the Gateway or GatewayClass you want to instrument.

Automatic configuration not detecting proxies

Symptom: No proxy configuration resources are created.

Solutions:

• Check that autoDetect is set to true or proxies are manually specified
• Verify the Cluster Agent logs for proxy detection messages
• Verify that your proxies are installed and have the expected Kubernetes resources (Gateway, GatewayClass)
• Try manually specifying proxy types using the proxies parameter

Proxy configuration not created

Symptom: The controller is running but configuration resources are missing.

Solutions:

• Check Cluster Agent logs for RBAC permission errors
• Verify the Cluster Agent service account has permissions to create the proxy configuration resources
• Verify that the processor service exists and is accessible
• Check for conflicting existing policies or filters

Traffic not being processed

Symptom: No security events appear in the Datadog UI.

Solutions:

• Verify the security processor deployment is running: kubectl get pods -n datadog -l app=datadog-aap-extproc
• Look for warning logs in your reverse proxies concerning this part of the configuration.
• Check processor logs for connection errors: kubectl logs -n datadog -l app=datadog-aap-extproc
• Verify the processor service is correctly configured and resolvable
• Test connectivity from gateway pods to the processor service
• Verify that Remote Configuration is enabled in your Datadog Agent

Security processor connection issues

Symptom: Gateways cannot reach the security processor.

Solutions:

• Verify the processor service name and namespace match your configuration
• Check for NetworkPolicy rules blocking cross-namespace traffic
• Test DNS resolution from gateway pods: nslookup datadog-aap-extproc-service.datadog.svc.cluster.local
• Verify the processor port configuration matches the service definition

RBAC permission errors

Symptom: Cluster Agent logs show permission denied errors.

Solutions:

• Verify the Cluster Agent ClusterRole includes permissions for:
  • gateway.networking.k8s.io/gateways
  • gateway.networking.k8s.io/gatewayclasses
• Check that the ClusterRoleBinding references the correct service account
• Make sure you are using the newest version of the Datadog Helm Chart or Operator.

Further Reading

Documentation, liens et articles supplémentaires utiles:

Collect your application tracesDOCUMENTATION Collect your application logsDOCUMENTATION App and API Protection for Envoy GatewayDOCUMENTATION App and API Protection for IstioDOCUMENTATION App and API Protection for ingress-nginxDOCUMENTATION OOTB App and API Protection RulesDOCUMENTATION Troubleshooting App and API ProtectionDOCUMENTATION