Monitor HAProxy on Kubernetes with OpenTelemetry

This Kubernetes integration automatically discovers HAProxy pods in your cluster and collects metrics without manual configuration for each instance. It uses the OpenTelemetry haproxyreceiver and receivercreator to dynamically monitor HAProxy sessions, requests, errors, and backend server health across your containerized environment.

Tip

Running HAProxy on Linux hosts? See the self-hosted installation guide for virtual machine or bare metal deployments.

Before you begin

Ensure you have:

Valid New Relic license key
HAProxy pods with the stats endpoint enabled (typically on port 8404)
A label on each HAProxy pod for discovery (for example, app: haproxy)
If you're using the Helm install method, complete the base Kubernetes OpenTelemetry Helm installation

Set up HAProxy monitoring

Add the following sections under deployment.extraConfig in your values.yaml:

Important

In deployment.extraConfig, pipelines: must be defined at the root level — not nested under service:. The Helm chart template maps extraConfig.pipelines into service.pipelines, and extraConfig.service.extensions into service.extensions.

Extensions:

extensions:
  k8s_observer:
    auth_type: serviceAccount
    observe_pods: true
    observe_nodes: false

Environment variables (add under deployment.envs):

deployment:
  envs:
    - name: OTEL_RESOURCE_ATTRIBUTES
      value: "host.id=$(KUBE_NODE_NAME)"

Important

The host.id resource attribute is required for entity synthesis. Container images don't have /etc/machine-id, so you must inject it via OTEL_RESOURCE_ATTRIBUTES. The Helm chart already sets KUBE_NODE_NAME from the Kubernetes Downward API.

Receivers:

receivers:
  receiver_creator/haproxy:
    watch_observers: [k8s_observer]
    receivers:
      haproxy:
        rule: type == "pod" && labels["app"] == "haproxy"
        config:
          endpoint: 'http://`endpoint`:8404/stats'
          collection_interval: 30s

Processors:

processors:
  resource/cluster:
    attributes:
      - key: k8s.cluster.name
        value: "<YOUR_CLUSTER_NAME>"
        action: insert

  transform/haproxy:
    metric_statements:
      - context: resource
        statements:
          - set(attributes["haproxy.display.name"], Concat([
              "server",
              "k8s",
              attributes["k8s.cluster.name"],
              attributes["k8s.namespace.name"],
              "pod",
              attributes["k8s.pod.name"],
              "haproxy",
              "8404"
            ], ":"))

  transform/metadata_nullify:
    metric_statements:
      - context: metric
        statements:
          - set(description, "")
          - set(unit, "")

Service extensions:

service:
  extensions: [k8s_observer]

Pipelines:

pipelines:
  metrics/haproxy:
    receivers: [receiver_creator/haproxy]
    processors: [batch, resource/cluster, transform/haproxy, transform/metadata_nullify, memory_limiter]
    exporters: [otlphttp/newrelic]

Configuration parameters

The following table describes the key configuration parameters:

Parameter	Description
`labels["app"] == "haproxy"`	Pod label selector used to discover HAProxy pods. Update the rule to match your pod labels.
`<YOUR_PORT_NUMBER>`	The port where HAProxy exposes its stats endpoint.
`<YOUR_CLUSTER_NAME>`	Replace with your Kubernetes cluster name for identification in New Relic.
`collection_interval`	How often to collect metrics. The default is `30s`.

Apply the updated values to your running Helm release:
bash
```
$helm upgrade nr-k8s-otel-collector newrelic/nr-k8s-otel-collector \
>  --namespace newrelic \
>  --reuse-values \
>  -f values.yaml
```

After completing the base Kubernetes OpenTelemetry manifest installation, configure HAProxy monitoring:

Update the collector image to use NRDOT collector.
In the deployment.yaml file in your local rendered directory, update the image to:
```
image: newrelic/nrdot-collector:latest
```
Mount the host's /etc/machine-id into the collector container by adding these to the deployment spec:
```
volumeMounts:
  - name: machine-id
    mountPath: /etc/machine-id
    readOnly: true
volumes:
  - name: machine-id
    hostPath:
      path: /etc/machine-id
      type: File
```
Important
The host.id resource attribute is required for entity synthesis. The resourcedetection processor reads /etc/machine-id to populate it. Without this volume mount, host.id won't be set and entities won't be created.

Replace the content in deployment-configmap.yaml under deployment-config.yaml: | with:

extensions:
  health_check:
    endpoint: 0.0.0.0:13133
  k8s_observer:
    auth_type: serviceAccount
    observe_pods: true
    observe_nodes: false

receivers:
  receiver_creator/haproxy:
    watch_observers: [k8s_observer]
    receivers:
      haproxy:
        rule: type == "pod" && labels["app"] == "haproxy"
        config:
          endpoint: 'http://`endpoint`:8404/stats'
          collection_interval: 30s

processors:
  resourcedetection:
    detectors: [env, system]
    system:
      resource_attributes:
        host.name:
          enabled: true
        host.id:
          enabled: true
        os.type:
          enabled: true

  batch:
    send_batch_max_size: 1000
    timeout: 30s
    send_batch_size: 800

  memory_limiter:
    check_interval: 1s
    limit_percentage: 80
    spike_limit_percentage: 25

  resource/cluster:
    attributes:
      - key: k8s.cluster.name
        value: "<YOUR_CLUSTER_NAME>"
        action: insert

  transform/haproxy:
    metric_statements:
      - context: resource
        statements:
          - set(attributes["haproxy.display.name"], Concat([
              "server",
              "k8s",
              attributes["k8s.cluster.name"],
              attributes["k8s.namespace.name"],
              "pod",
              attributes["k8s.pod.name"],
              "haproxy",
              "8404"
            ], ":"))

  transform/metadata_nullify:
    metric_statements:
      - context: metric
        statements:
          - set(description, "")
          - set(unit, "")

exporters:
  otlp_http/newrelic:
    endpoint: "<YOUR_NEWRELIC_OTLP_ENDPOINT>"
    headers:
      api-key: ${env:NR_LICENSE_KEY}

service:
  extensions: [health_check, k8s_observer]
  pipelines:
    metrics/haproxy:
      receivers: [receiver_creator/haproxy]
      processors: [resourcedetection, batch, resource/cluster, transform/haproxy, transform/metadata_nullify, memory_limiter]
      exporters: [otlp_http/newrelic]

Delete the DaemonSet and kube-state-metrics deployment (not needed for HAProxy-only monitoring):

bash

$kubectl delete daemonset nr-k8s-otel-collector-daemonset -n newrelic
$kubectl delete deployment nr-k8s-otel-collector-kube-state-metrics -n newrelic

Restart the collector deployment to apply your changes:

bash

$kubectl rollout restart deployment nr-k8s-otel-collector-deployment -n newrelic

Configuration parameters

Parameter	Description
`labels["app"] == "haproxy"`	Pod label selector used to discover HAProxy pods. Update the rule to match your pod labels.
`<YOUR_PORT_NUMBER>`	The port where HAProxy exposes its stats endpoint.
`<YOUR_CLUSTER_NAME>`	Replace with your Kubernetes cluster name for identification in New Relic.
`<YOUR_NEWRELIC_OTLP_ENDPOINT>`	Your New Relic OTLP endpoint. Use `https://otlp.nr-data.net` for US, `https://otlp.eu01.nr-data.net` for EU.

Deploy the OpenTelemetry Collector Contrib to your Kubernetes cluster. The collector automatically discovers and scrapes metrics from your HAProxy pods. Choose between Helm (using the upstream OTel Helm chart) or raw Kubernetes manifests.

Create a values.yaml file with the following HAProxy monitoring configuration:

mode: deployment

image:
  repository: otel/opentelemetry-collector-contrib

clusterRole:
  create: true
  rules:
    - apiGroups: [""]
      resources: ["pods", "nodes"]
      verbs: ["get", "list", "watch"]
    - apiGroups: ["apps"]
      resources: ["replicasets"]
      verbs: ["get", "list", "watch"]

extraVolumes:
  - name: machine-id
    hostPath:
      path: /etc/machine-id
      type: File

extraVolumeMounts:
  - name: machine-id
    mountPath: /etc/machine-id
    readOnly: true

config:
  extensions:
    health_check:
      endpoint: ${env:MY_POD_IP}:13133
    k8s_observer:
      auth_type: serviceAccount
      observe_pods: true
      observe_nodes: false

  receivers:
    receiver_creator/haproxy:
      watch_observers: [k8s_observer]
      receivers:
        haproxy:
          rule: type == "pod" && labels["app"] == "haproxy"
          config:
            endpoint: 'http://`endpoint`:8404/stats'
            collection_interval: 30s

  processors:
    resourcedetection:
      detectors: [env, system]
      system:
        resource_attributes:
          host.name:
            enabled: true
          host.id:
            enabled: true
          os.type:
            enabled: true

    batch:
      send_batch_size: 1024
      timeout: 30s

    memory_limiter:
      check_interval: 1s
      limit_percentage: 80
      spike_limit_percentage: 25

    resource/cluster:
      attributes:
        - key: k8s.cluster.name
          value: "<YOUR_CLUSTER_NAME>"
          action: insert

    transform/haproxy:
      metric_statements:
        - context: resource
          statements:
            - set(attributes["haproxy.display.name"], Concat(["server", "k8s", attributes["k8s.cluster.name"], attributes["k8s.namespace.name"], "pod", attributes["k8s.pod.name"], "haproxy", "8404"], ":"))

    transform/metadata_nullify:
      metric_statements:
        - context: metric
          statements:
            - set(description, "")
            - set(unit, "")

  exporters:
    otlp_http/newrelic:
      endpoint: "<YOUR_NEWRELIC_OTLP_ENDPOINT>"
      headers:
        api-key: "<YOUR_NEW_RELIC_LICENSE_KEY>"

  service:
    extensions: [health_check, k8s_observer]
    pipelines:
      metrics/haproxy:
        receivers: [receiver_creator/haproxy]
        processors: [resourcedetection, batch, resource/cluster, transform/haproxy, transform/metadata_nullify, memory_limiter]
        exporters: [otlp_http/newrelic]

Configuration parameters

Parameter	Description
`labels["app"] == "haproxy"`	Pod label selector used to discover HAProxy pods. Update the rule to match your pod labels.
`<YOUR_PORT_NUMBER>`	The port where HAProxy exposes its stats endpoint.
`<YOUR_CLUSTER_NAME>`	Replace with your Kubernetes cluster name for identification in New Relic.
`<YOUR_NEWRELIC_OTLP_ENDPOINT>`	Your New Relic OTLP endpoint. Use `https://otlp.nr-data.net` for US, `https://otlp.eu01.nr-data.net` for EU.
`<YOUR_NEW_RELIC_LICENSE_KEY>`	Your New Relic ingest license key.

Install using the upstream OpenTelemetry Helm chart:

bash

$helm repo add open-telemetry https://open-telemetry.github.io/opentelemetry-helm-charts
$helm repo update
$helm upgrade --install haproxy-otel-collector open-telemetry/opentelemetry-collector \
>  -f values.yaml \
>  -n newrelic --create-namespace

Use this option if you can't use Helm or prefer to manage Kubernetes resources directly with kubectl.

Create a secret with your New Relic license key:

bash

$kubectl create namespace newrelic
$kubectl create secret generic otel-collector-haproxy-secret \
>  --namespace newrelic \
>  --from-literal=licenseKey='<YOUR_NEW_RELIC_LICENSE_KEY>'

Save the following manifest to a file (for example, otel-collector-haproxy.yaml):

---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: otel-collector-haproxy
  namespace: newrelic
  labels:
    app: otel-collector-haproxy
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: otel-collector-haproxy
  labels:
    app: otel-collector-haproxy
rules:
  - apiGroups: [""]
    resources: ["pods", "nodes"]
    verbs: ["get", "list", "watch"]
  - apiGroups: ["apps"]
    resources: ["replicasets"]
    verbs: ["get", "list", "watch"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: otel-collector-haproxy
  labels:
    app: otel-collector-haproxy
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: otel-collector-haproxy
subjects:
  - kind: ServiceAccount
    name: otel-collector-haproxy
    namespace: newrelic
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: otel-collector-haproxy-config
  namespace: newrelic
  labels:
    app: otel-collector-haproxy
data:
  collector-config.yaml: |
    extensions:
      health_check:
        endpoint: 0.0.0.0:13133
      k8s_observer:
        auth_type: serviceAccount
        observe_pods: true
        observe_nodes: false

    receivers:
      receiver_creator/haproxy:
        watch_observers: [k8s_observer]
        receivers:
          haproxy:
            rule: type == "pod" && labels["app"] == "haproxy"
            config:
              endpoint: 'http://`endpoint`:8404/stats'
              collection_interval: 30s

    processors:
      resourcedetection:
        detectors: [env, system]
        system:
          resource_attributes:
            host.name:
              enabled: true
            host.id:
              enabled: true
            os.type:
              enabled: true

      batch:
        send_batch_size: 1024
        timeout: 30s

      memory_limiter:
        check_interval: 1s
        limit_percentage: 80
        spike_limit_percentage: 25

      resource/cluster:
        attributes:
          - key: k8s.cluster.name
            value: "<YOUR_CLUSTER_NAME>"
            action: insert

      transform/haproxy:
        metric_statements:
          - context: resource
            statements:
              - set(attributes["haproxy.display.name"], Concat(["server", "k8s", attributes["k8s.cluster.name"], attributes["k8s.namespace.name"], "pod", attributes["k8s.pod.name"], "haproxy", "8404"], ":"))

      transform/metadata_nullify:
        metric_statements:
          - context: metric
            statements:
              - set(description, "")
              - set(unit, "")

    exporters:
      otlp_http/newrelic:
        endpoint: "<YOUR_NEWRELIC_OTLP_ENDPOINT>"
        headers:
          api-key: ${env:NR_LICENSE_KEY}

    service:
      extensions: [health_check, k8s_observer]
      pipelines:
        metrics/haproxy:
          receivers: [receiver_creator/haproxy]
          processors: [resourcedetection, batch, resource/cluster, transform/haproxy, transform/metadata_nullify, memory_limiter]
          exporters: [otlp_http/newrelic]
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: otel-collector-haproxy
  namespace: newrelic
  labels:
    app: otel-collector-haproxy
spec:
  replicas: 1
  selector:
    matchLabels:
      app: otel-collector-haproxy
  template:
    metadata:
      labels:
        app: otel-collector-haproxy
    spec:
      serviceAccountName: otel-collector-haproxy
      containers:
      - name: otel-collector
        image: otel/opentelemetry-collector-contrib:latest
        command:
        - "/otelcol-contrib"
        - "--config=/conf/collector-config.yaml"
        env:
        - name: NR_LICENSE_KEY
          valueFrom:
            secretKeyRef:
              name: otel-collector-haproxy-secret
              key: licenseKey
        - name: GOGC
          value: "80"
        ports:
        - name: health
          containerPort: 13133
          protocol: TCP
        resources:
          limits:
            cpu: 500m
            memory: 512Mi
          requests:
            cpu: 100m
            memory: 256Mi
        volumeMounts:
        - name: config
          mountPath: /conf
        - name: machine-id
          mountPath: /etc/machine-id
          readOnly: true
      volumes:
      - name: config
        configMap:
          name: otel-collector-haproxy-config
          items:
          - key: collector-config.yaml
            path: collector-config.yaml
      - name: machine-id
        hostPath:
          path: /etc/machine-id
          type: File

Important

The /etc/machine-id volume mount is required for entity synthesis. The resourcedetection processor reads this file to populate the host.id resource attribute. Without it, entities won't be created in New Relic.

Apply the manifest:

bash

$kubectl apply -f otel-collector-haproxy.yaml

Verify the collector is running and discovering HAProxy pods:

bash

$kubectl get pods -n newrelic -l app=otel-collector-haproxy
$kubectl logs -n newrelic -l app=otel-collector-haproxy

You should see log lines showing starting receiver with your HAProxy pod IPs.

Configuration parameters

Parameter	Description
`labels["app"] == "haproxy"`	Pod label selector used to discover HAProxy pods. Update the rule to match your pod labels.
`<YOUR_PORT_NUMBER>`	The port where HAProxy exposes its stats endpoint.
`<YOUR_CLUSTER_NAME>`	Replace with your Kubernetes cluster name for identification in New Relic.
`<YOUR_NEWRELIC_OTLP_ENDPOINT>`	Your New Relic OTLP endpoint. Use `https://otlp.nr-data.net` for US, `https://otlp.eu01.nr-data.net` for EU.
`<YOUR_NEW_RELIC_LICENSE_KEY>`	Your New Relic ingest license key.

Use this approach if your HAProxy pods already expose a Prometheus /metrics endpoint (HAProxy 2.0+) and you prefer to scrape Prometheus metrics rather than connect to the CSV stats endpoint.

Tip

If you don't already have the Prometheus endpoint enabled, use the NRDOT collector or OpenTelemetry Collector Contrib tabs instead. They connect directly to the HAProxy stats endpoint without needing Prometheus support.

Important

This approach requires HAProxy 2.0+ with the built-in Prometheus endpoint enabled. Your HAProxy configuration must include:

frontend stats
  bind *:8404
  http-request use-service prometheus-exporter if { path /metrics }

And your HAProxy pods must expose a named port for the metrics endpoint:

ports:
  - name: metrics
    containerPort: 8404

Complete the base Kubernetes OpenTelemetry Helm installation for nr-k8s-otel-collector before proceeding.

Add the following sections under deployment.configMap.extraConfig in your values.yaml:

Important

In deployment.configMap.extraConfig, pipelines: must be defined at the root level — not nested under service:. The Helm chart template maps extraConfig.pipelines into service.pipelines, and extraConfig.service.extensions into service.extensions.

Extensions:

extensions:
  k8s_observer:
    auth_type: serviceAccount
    observe_pods: true
    observe_nodes: false

Environment variables (add under deployment.envs):

deployment:
  envs:
    - name: OTEL_RESOURCE_ATTRIBUTES
      value: "host.id=$(KUBE_NODE_NAME)"

Important

The host.id resource attribute is required for entity synthesis. The resourcedetection processor reads OTEL_RESOURCE_ATTRIBUTES to populate it. The Helm chart already sets KUBE_NODE_NAME from the Kubernetes Downward API.

Receivers:

receivers:
  receiver_creator/haproxy_prom:
    watch_observers: [k8s_observer]
    receivers:
      prometheus:
        rule: type == "port" && pod.labels["app"] == "haproxy" && port == 8404
        config:
          config:
            scrape_configs:
              - job_name: haproxy_prom
                metrics_path: /metrics
                scrape_interval: 30s
                static_configs:
                  - targets:
                      - '`endpoint`'
        resource_attributes:
          haproxy.addr: 'http://`endpoint`/metrics'

Processors:

processors:
  resourcedetection:
    detectors: [env]

  filter/haproxy:
    metrics:
      include:
        match_type: regexp
        metric_names:
          - "haproxy_frontend_bytes_in_total"
          - "haproxy_frontend_bytes_out_total"
          - "haproxy_frontend_requests_denied_total"
          - "haproxy_frontend_request_errors_total"
          - "haproxy_frontend_http_requests_total"
          - "haproxy_frontend_current_sessions"
          - "haproxy_frontend_max_session_rate"
          - "haproxy_frontend_http_requests_rate_max"
          - "haproxy_frontend_connections_rate_max"
          - "haproxy_backend_bytes_in_total"
          - "haproxy_backend_bytes_out_total"
          - "haproxy_backend_connection_errors_total"
          - "haproxy_backend_http_requests_total"
          - "haproxy_backend_current_sessions"
          - "haproxy_backend_max_session_rate"
          - "haproxy_backend_current_queue"
          - "haproxy_server_bytes_in_total"
          - "haproxy_server_bytes_out_total"
          - "haproxy_server_connection_errors_total"
          - "haproxy_server_current_sessions"
          - "haproxy_server_max_session_rate"
          - "haproxy_server_current_queue"

  transform/add_labels:
    metric_statements:
      - context: datapoint
        statements:
          - set(attributes["haproxy.service_name"], "FRONTEND") where IsMatch(metric.name, "^haproxy_frontend_.*")
          - set(attributes["haproxy.service_name"], "BACKEND") where IsMatch(metric.name, "^haproxy_backend_.*")
          - set(attributes["haproxy.service_name"], attributes["server"]) where IsMatch(metric.name, "^haproxy_server_.*")
          - set(attributes["haproxy.proxy_name"], attributes["proxy"]) where attributes["proxy"] != nil
          - delete_key(attributes, "proxy") where attributes["proxy"] != nil
          - delete_key(attributes, "server") where attributes["server"] != nil

  metricstransform/rename:
    transforms:
      - include: haproxy_frontend_bytes_in_total
        action: update
        new_name: haproxy.bytes.input
      - include: haproxy_backend_bytes_in_total
        action: update
        new_name: haproxy.bytes.input
      - include: haproxy_server_bytes_in_total
        action: update
        new_name: haproxy.bytes.input
      - include: haproxy_frontend_bytes_out_total
        action: update
        new_name: haproxy.bytes.output
      - include: haproxy_backend_bytes_out_total
        action: update
        new_name: haproxy.bytes.output
      - include: haproxy_server_bytes_out_total
        action: update
        new_name: haproxy.bytes.output
      - include: haproxy_backend_connection_errors_total
        action: update
        new_name: haproxy.connections.errors
      - include: haproxy_server_connection_errors_total
        action: update
        new_name: haproxy.connections.errors
      - include: haproxy_frontend_requests_denied_total
        action: update
        new_name: haproxy.requests.denied
      - include: haproxy_frontend_request_errors_total
        action: update
        new_name: haproxy.requests.errors
      - include: haproxy_frontend_http_requests_total
        action: update
        new_name: haproxy.requests.total
      - include: haproxy_backend_http_requests_total
        action: update
        new_name: haproxy.requests.total
      - include: haproxy_backend_current_queue
        action: update
        new_name: haproxy.requests.queued
      - include: haproxy_server_current_queue
        action: update
        new_name: haproxy.requests.queued
      - include: haproxy_frontend_current_sessions
        action: update
        new_name: haproxy.sessions.count
      - include: haproxy_backend_current_sessions
        action: update
        new_name: haproxy.sessions.count
      - include: haproxy_server_current_sessions
        action: update
        new_name: haproxy.sessions.count
      - include: haproxy_frontend_http_requests_rate_max
        action: update
        new_name: haproxy.requests.rate
      - include: haproxy_frontend_connections_rate_max
        action: update
        new_name: haproxy.connections.rate
      - include: haproxy_frontend_max_session_rate
        action: update
        new_name: haproxy.sessions.rate
      - include: haproxy_backend_max_session_rate
        action: update
        new_name: haproxy.sessions.rate
      - include: haproxy_server_max_session_rate
        action: update
        new_name: haproxy.sessions.rate

  resource/cluster:
    attributes:
      - key: k8s.cluster.name
        value: "<YOUR_CLUSTER_NAME>"
        action: insert

  transform/haproxy:
    metric_statements:
      - context: resource
        statements:
          - set(attributes["haproxy.display.name"], Concat(["server", "k8s", attributes["k8s.cluster.name"], attributes["k8s.namespace.name"], "pod", attributes["k8s.pod.name"], "haproxy", "8404"], ":"))

  transform/metadata_nullify:
    metric_statements:
      - context: metric
        statements:
          - set(description, "")
          - set(unit, "")

Service extensions:

service:
  extensions: [k8s_observer]

Pipelines:

pipelines:
  metrics/haproxy_prom:
    receivers: [receiver_creator/haproxy_prom]
    processors: [resourcedetection, filter/haproxy, transform/add_labels, metricstransform/rename, resource/cluster, batch, transform/haproxy, transform/metadata_nullify, memory_limiter]
    exporters: [otlp_http/newrelic]

Configuration parameters

Parameter	Description
`labels["app"] == "haproxy"`	Pod label selector used to discover HAProxy pods. Update the rule to match your pod labels.
`<YOUR_PORT_NUMBER>`	The port where HAProxy exposes its Prometheus endpoint.
`<YOUR_CLUSTER_NAME>`	Replace with your Kubernetes cluster name for identification in New Relic.
`scrape_interval`	How often to scrape Prometheus metrics. Default: `30s`.

Apply the updated values to your running Helm release:

bash

$helm upgrade nr-k8s-otel-collector newrelic/nr-k8s-otel-collector \
>  --namespace newrelic \
>  --reuse-values \
>  -f values.yaml

Create a values.yaml file with the Prometheus receiver configuration:

mode: deployment

image:
  repository: otel/opentelemetry-collector-contrib

clusterRole:
  create: true
  rules:
    - apiGroups: [""]
      resources: ["pods", "nodes"]
      verbs: ["get", "list", "watch"]
    - apiGroups: ["apps"]
      resources: ["replicasets"]
      verbs: ["get", "list", "watch"]

extraVolumes:
  - name: machine-id
    hostPath:
      path: /etc/machine-id
      type: File

extraVolumeMounts:
  - name: machine-id
    mountPath: /etc/machine-id
    readOnly: true

config:
  extensions:
    health_check:
      endpoint: ${env:MY_POD_IP}:13133
    k8s_observer:
      auth_type: serviceAccount
      observe_pods: true
      observe_nodes: false

  receivers:
    receiver_creator/haproxy_prom:
      watch_observers: [k8s_observer]
      receivers:
        prometheus:
          rule: type == "port" && pod.labels["app"] == "haproxy" && port == 8404
          config:
            config:
              scrape_configs:
                - job_name: haproxy_prom
                  metrics_path: /metrics
                  scrape_interval: 30s
                  static_configs:
                    - targets:
                        - '`endpoint`'
          resource_attributes:
            haproxy.addr: 'http://`endpoint`/metrics'

  processors:
    resourcedetection:
      detectors: [env, system]
      system:
        resource_attributes:
          host.name:
            enabled: true
          host.id:
            enabled: true
          os.type:
            enabled: true

    filter/haproxy:
      metrics:
        include:
          match_type: regexp
          metric_names:
            - "haproxy_frontend_bytes_in_total"
            - "haproxy_frontend_bytes_out_total"
            - "haproxy_frontend_requests_denied_total"
            - "haproxy_frontend_request_errors_total"
            - "haproxy_frontend_http_requests_total"
            - "haproxy_frontend_current_sessions"
            - "haproxy_frontend_max_session_rate"
            - "haproxy_backend_bytes_in_total"
            - "haproxy_backend_bytes_out_total"
            - "haproxy_backend_connection_errors_total"
            - "haproxy_backend_http_requests_total"
            - "haproxy_backend_current_sessions"
            - "haproxy_backend_max_session_rate"
            - "haproxy_backend_current_queue"
            - "haproxy_server_bytes_in_total"
            - "haproxy_server_bytes_out_total"
            - "haproxy_server_connection_errors_total"
            - "haproxy_server_current_sessions"
            - "haproxy_server_max_session_rate"
            - "haproxy_server_current_queue"

    transform/add_labels:
      metric_statements:
        - context: datapoint
          statements:
            - set(attributes["haproxy.service_name"], "FRONTEND") where IsMatch(metric.name, "^haproxy_frontend_.*")
            - set(attributes["haproxy.service_name"], "BACKEND") where IsMatch(metric.name, "^haproxy_backend_.*")
            - set(attributes["haproxy.service_name"], attributes["server"]) where IsMatch(metric.name, "^haproxy_server_.*")
            - set(attributes["haproxy.proxy_name"], attributes["proxy"]) where attributes["proxy"] != nil
            - delete_key(attributes, "proxy") where attributes["proxy"] != nil
            - delete_key(attributes, "server") where attributes["server"] != nil

    metricstransform/rename:
      transforms:
        - include: haproxy_frontend_bytes_in_total
          action: update
          new_name: haproxy.bytes.input
        - include: haproxy_backend_bytes_in_total
          action: update
          new_name: haproxy.bytes.input
        - include: haproxy_server_bytes_in_total
          action: update
          new_name: haproxy.bytes.input
        - include: haproxy_frontend_bytes_out_total
          action: update
          new_name: haproxy.bytes.output
        - include: haproxy_backend_bytes_out_total
          action: update
          new_name: haproxy.bytes.output
        - include: haproxy_server_bytes_out_total
          action: update
          new_name: haproxy.bytes.output
        - include: haproxy_backend_connection_errors_total
          action: update
          new_name: haproxy.connections.errors
        - include: haproxy_server_connection_errors_total
          action: update
          new_name: haproxy.connections.errors
        - include: haproxy_frontend_requests_denied_total
          action: update
          new_name: haproxy.requests.denied
        - include: haproxy_frontend_request_errors_total
          action: update
          new_name: haproxy.requests.errors
        - include: haproxy_frontend_http_requests_total
          action: update
          new_name: haproxy.requests.total
        - include: haproxy_backend_http_requests_total
          action: update
          new_name: haproxy.requests.total
        - include: haproxy_backend_current_queue
          action: update
          new_name: haproxy.requests.queued
        - include: haproxy_server_current_queue
          action: update
          new_name: haproxy.requests.queued
        - include: haproxy_frontend_current_sessions
          action: update
          new_name: haproxy.sessions.count
        - include: haproxy_backend_current_sessions
          action: update
          new_name: haproxy.sessions.count
        - include: haproxy_server_current_sessions
          action: update
          new_name: haproxy.sessions.count
        - include: haproxy_frontend_max_session_rate
          action: update
          new_name: haproxy.sessions.rate
        - include: haproxy_backend_max_session_rate
          action: update
          new_name: haproxy.sessions.rate
        - include: haproxy_server_max_session_rate
          action: update
          new_name: haproxy.sessions.rate

    resource/cluster:
      attributes:
        - key: k8s.cluster.name
          value: "<YOUR_CLUSTER_NAME>"
          action: insert

    batch:
      send_batch_size: 1024
      timeout: 30s

    memory_limiter:
      check_interval: 1s
      limit_percentage: 80
      spike_limit_percentage: 25

    transform/haproxy:
      metric_statements:
        - context: resource
          statements:
            - set(attributes["haproxy.display.name"], Concat(["server", "k8s", attributes["k8s.cluster.name"], attributes["k8s.namespace.name"], "pod", attributes["k8s.pod.name"], "haproxy", "8404"], ":"))

    transform/metadata_nullify:
      metric_statements:
        - context: metric
          statements:
            - set(description, "")
            - set(unit, "")

  exporters:
    otlp_http/newrelic:
      endpoint: "<YOUR_NEWRELIC_OTLP_ENDPOINT>"
      headers:
        api-key: "<YOUR_NEW_RELIC_LICENSE_KEY>"

  service:
    extensions: [health_check, k8s_observer]
    pipelines:
      metrics/haproxy_prom:
        receivers: [receiver_creator/haproxy_prom]
        processors: [resourcedetection, filter/haproxy, transform/add_labels, metricstransform/rename, resource/cluster, batch, transform/haproxy, transform/metadata_nullify, memory_limiter]
        exporters: [otlp_http/newrelic]

Important

Configuration parameters

Parameter	Description
`labels["app"] == "haproxy"`	Pod label selector used to discover HAProxy pods. Update the rule to match your pod labels.
`<YOUR_PORT_NUMBER>`	The port where HAProxy exposes its Prometheus endpoint.
`<YOUR_CLUSTER_NAME>`	Replace with your Kubernetes cluster name for identification in New Relic.
`<YOUR_NEWRELIC_OTLP_ENDPOINT>`	Your New Relic OTLP endpoint. Use `https://otlp.nr-data.net` for US, `https://otlp.eu01.nr-data.net` for EU.
`<YOUR_NEW_RELIC_LICENSE_KEY>`	Your New Relic ingest license key.
`scrape_interval`	How often to scrape Prometheus metrics. Default: `30s`.

Install using the upstream OpenTelemetry Helm chart:

bash

$helm repo add open-telemetry https://open-telemetry.github.io/opentelemetry-helm-charts
$helm repo update
$helm upgrade --install haproxy-prom-collector open-telemetry/opentelemetry-collector \
>  -f values.yaml \
>  -n newrelic --create-namespace

View your data in New Relic

After a few minutes, HAProxy metrics appear in New Relic. To find your data:

Go to one.newrelic.com > All capabilities > All entities.
Search for your HAProxy entities by name or filter by entity type HAPROXYINSTANCE.
Select an entity to view the summary page with golden metrics (sessions per second, requests per second, connection errors).

Troubleshooting

Verify HAProxy pods have the correct label (for example, app: haproxy): kubectl get pods -l app=haproxy
Check that the stats endpoint responds: kubectl exec -it <haproxy-pod> -- curl -s http://localhost:8404/stats
Review collector logs for discovery errors: kubectl logs -n newrelic -l app.kubernetes.io/name=nr-k8s-otel-collector -f (NRDOT) or kubectl logs -n newrelic -l app=otel-collector-haproxy -f (OTel Contrib manifest)
Ensure the collector ServiceAccount has get, list, and watch permissions on pods.

Each HAProxy pod should create a separate entity in New Relic. If you see only one entity:

Verify the collector has host.id set. In containerized environments, add this to the collector deployment:

env:
  - name: K8S_NODE_NAME
    valueFrom:
      fieldRef:
        fieldPath: spec.nodeName
  - name: OTEL_RESOURCE_ATTRIBUTES
    value: "host.id=$(K8S_NODE_NAME)"

Confirm different haproxy.addr values per pod by querying: FROM Metric SELECT uniques(haproxy.addr) WHERE metricName LIKE 'haproxy.%' SINCE 10 minutes ago

Next steps

Review the metrics reference for a complete list of collected metrics.
Learn how to find and query your HAProxy data.
Set up alerts based on HAProxy golden metrics.

Monitor HAProxy on Kubernetes with OpenTelemetry

Tip

Before you begin

Set up HAProxy monitoring

Helm install

Important

Important

Configuration parameters

Manifest install

Important

Configuration parameters

Helm install

Configuration parameters

Manifest install

Important

Configuration parameters

Tip

Important

NRDOT collector — Helm install

Important

Important

Configuration parameters

OTel Collector Contrib — Helm install

Important

Configuration parameters

View your data in New Relic

Troubleshooting

No data appears in New Relic

Multiple HAProxy pods show as one entity

Collector can't reach HAProxy stats endpoint

Next steps

Monitor HAProxy on Kubernetes with OpenTelemetry

Tip

Before you begin .css-21sua1{background:none;border:none;width:0;padding:0;}

Set up HAProxy monitoring

Manifest install

Helm install

Manifest install

Tip

Important

NRDOT collector — Helm install

OTel Collector Contrib — Helm install

View your data in New Relic

Troubleshooting

No data appears in New Relic

Multiple HAProxy pods show as one entity

Collector can't reach HAProxy stats endpoint

Next steps

Before you begin