<br />

This document describes how to configure your Google Kubernetes Engine deployment
so that you can use Google Cloud Managed Service for Prometheus to collect metrics from

GKE Inference Gateway. This document shows you how to do the following:

- Set up GKE Inference Gateway to report metrics.
- Access a predefined dashboard in Cloud Monitoring to view the metrics.

<br />

These instructions apply only if you are using [managed collection](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/setup-managed.md.txt)
with Managed Service for Prometheus.
If you are using self-deployed collection, then see the

[GKE Inference Gateway documentation](https://github.com/kubernetes-sigs/gateway-api-inference-extension/blob/2d7613c9b2bebada4c212495326272137c5846db/site-src/guides/metrics-and-observability.md)

for installation information.

These instructions are provided as an example and are expected to work in
most Kubernetes environments.

If you are having trouble installing an
application or exporter due to restrictive security or organizational policies,
then we recommend you consult open-source documentation for support.

For information about GKE Inference Gateway, see [GKE Inference Gateway](https://github.com/kubernetes-sigs/gateway-api-inference-extension/).

## Prerequisites

To collect metrics from

the GKE Inference Gateway exporter

by using
Managed Service for Prometheus and managed collection, your deployment must
meet the following requirements:

- Your cluster must be running Google Kubernetes Engine version 1.28.15-gke.2475000 or later.
- You must be running Managed Service for Prometheus with managed collection enabled. For more information, see [Get started with managed collection](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/setup-managed.md.txt).

  <br />

<br />

> [!NOTE]
> **Note:** This exporter and the **GKE Inference Gateway Prometheus Overview** dashboard target legacy Gateway API Inference Extension (v1.5.0 and earlier) and llm-d (earlier than v0.8.0) deployments that emit `inferenceobjective*` metrics. For [llm-d](https://llm-d.ai/) version v0.8.0 and later (both Standalone Mode and Gateway Mode with GKE Inference Gateway), use the [llm-d exporter
> and unified **llm-d Overview** dashboard](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/exporters/llm-d) instead.

<br />

The GKE Inference Gateway exporter exposes Prometheus-format metrics
automatically; you don't have to install it separately.

To verify that the GKE Inference Gateway exporter is emitting
metrics on the expected endpoints, do the following:

1. Add a secret, ServiceAccount, ClusterRole, and ClusterBinding.
   The GKE Inference Gateway exporter observability endpoints are protected by
   the auth token. To obtain credentials, the client requires a Secret that
   maps to a service account with the connected ClusterRole, for the
   `nonResourceURLs: "/metrics", verbs: get` rule.
   For more information, see
   [Create a secret for a service account](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/exporters/inference-optimized-gateway.md.txt#exporter-secret).

2. Set up port forwarding by using the following command:

   ```
   kubectl -n NAMESPACE_NAME port-forward POD_NAME 9090
   ```
3. In another window, do the following:

   1. Fetch the token by running the following command:

      ```
      TOKEN=$(kubectl -n default get secret inference-gateway-sa-metrics-reader-secret  -o jsonpath='{.secrets[0].name}' -o jsonpath='{.data.token}' | base64 --decode)
      ```
   2. Access the endpoint `localhost:9090/metrics` using the `curl` utility:

      ```
      curl -H "Authorization: Bearer $TOKEN" localhost:9090/metrics
      ```

## Create a secret for a service account

For the protected the GKE Inference Gateway exporter endpoint, the
Managed Service for Prometheus Operator requires a secret for
authorized metric collection in the `gmp-system` namespace.

If your cluster is using Autopilot mode, then replace
`gmp-system` with `gke-gmp-system`.

You can use the following Secret, ServiceAccount, ClusterRole and ClusterRoleBinding configuration:

```yaml
# Copyright 2025 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: inference-gateway-metrics-reader
rules:
- nonResourceURLs:
  - /metrics
  verbs:
  - get
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: inference-gateway-sa-metrics-reader
  namespace: default
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: inference-gateway-sa-metrics-reader-role-binding
  namespace: default
subjects:
- kind: ServiceAccount
  name: inference-gateway-sa-metrics-reader
  namespace: default
roleRef:
  kind: ClusterRole
  name: inference-gateway-metrics-reader
  apiGroup: rbac.authorization.k8s.io
---
apiVersion: v1
kind: Secret
metadata:
  name: inference-gateway-sa-metrics-reader-secret
  namespace: default
  annotations:
    kubernetes.io/service-account.name: inference-gateway-sa-metrics-reader
type: kubernetes.io/service-account-token
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: inference-gateway-sa-metrics-reader-secret-read
rules:
- resources:
  - secrets
  apiGroups: [""]
  verbs: ["get", "list", "watch"]
  resourceNames: ["inference-gateway-sa-metrics-reader-secret"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: gmp-system:collector:inference-gateway-sa-metrics-reader-secret-read
  namespace: default
roleRef:
  name: inference-gateway-sa-metrics-reader-secret-read
  kind: ClusterRole
  apiGroup: rbac.authorization.k8s.io
subjects:
- name: collector
  namespace: gmp-system
  kind: ServiceAccount
```

For more information, see the exporter's [Metric \& Observability
guide](https://github.com/kubernetes-sigs/gateway-api-inference-extension/blob/2d7613c9b2bebada4c212495326272137c5846db/site-src/guides/metrics-and-observability.md#scrape-metrics--pprof-profiles).


To apply configuration changes from a local file, run the following command:

```
kubectl apply -n NAMESPACE_NAME -f FILE_NAME
```

<br />

You can also
[use Terraform](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/setup-managed.md.txt#terraform-scrape)
to manage your configurations.

## Define a ClusterPodMonitoring resource

For target discovery, the Managed Service for Prometheus Operator
requires a ClusterPodMonitoring resource that corresponds to
the GKE Inference Gateway exporter in the same namespace.

You can use the following ClusterPodMonitoring configuration:

```yaml
# Copyright 2025 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

apiVersion: monitoring.googleapis.com/v1
kind: ClusterPodMonitoring
metadata:
  name: inference-optimized-gateway-monitoring
  labels:
    app.kubernetes.io/name: inference-optimized-gateway
    app.kubernetes.io/part-of: google-cloud-managed-prometheus
spec:
  endpoints:
  - port: metrics
    scheme: http
    interval: 5s
    path: /metrics
    authorization:
      type: Bearer
      credentials:
        secret:
          name: inference-gateway-sa-metrics-reader-secret
          key: token
          namespace: default
  selector:
    matchLabels:
      app: inference-gateway-ext-proc
```

GKE Inference Gateway uses the `ClusterPodMonitoring` resource instead of the
`PodMonitoring` resource because it needs to access the secret from another
namespace.

In the `matchLabels` selector of the `ClusterPodMonitoring` configuration,
you can replace the `app` value of `inference-gateway-ext-proc` with labels
from your GKE Inference Gateway deployment.


Ensure that the values of the `port` and `matchLabels` fields match those of the GKE Inference Gateway pods you want to monitor.

To apply configuration changes from a local file, run the following command:

```
kubectl apply -n NAMESPACE_NAME -f FILE_NAME
```

<br />

You can also
[use Terraform](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/setup-managed.md.txt#terraform-scrape)
to manage your configurations.

## Verify the configuration

You can use Metrics Explorer to verify that you correctly configured
the GKE Inference Gateway exporter. It might take one or two minutes for
Cloud Monitoring to ingest your metrics.

To verify the metrics are ingested, do the following:

1. In the Google Cloud console, go to the
   **Metrics explorer** page:

   [Go to **Metrics explorer**](https://console.cloud.google.com/monitoring/metrics-explorer)

   <br />

   If you use the search bar to find this page, then select the result whose subheading is
   **Monitoring**.
2. In the toolbar of the query-builder pane, select the button whose name is **PromQL**.
3. Enter and run the following query:  

   ```
   inference_model_request_total{cluster="CLUSTER_NAME", namespace="NAMESPACE_NAME"}
   ```

<br />

## View dashboards

The Cloud Monitoring integration includes

the **GKE Inference Gateway Prometheus Overview** dashboard.

Dashboards are automatically installed when you configure the integration.
You can also view static previews of dashboards without installing the
integration.


To view an installed dashboard, do the following:

1. In the Google Cloud console, go to the **Dashboards** page:

   [Go to **Dashboards**](https://console.cloud.google.com/monitoring/dashboards)

   <br />

   If you use the search bar to find this page, then select the result whose subheading is
   **Monitoring**.
2. Select the **Dashboard List** tab.
3. Choose the **Integrations** category.
4. Click the name of the dashboard, for example, **GKE Inference Gateway Prometheus Overview**.

<br />

To view a static preview of the dashboard, do the following:

1. In the Google Cloud console, go to the
   ![](https://docs.cloud.google.com/static/monitoring/images/integrations-icon.png) **Integrations**
   page:

   [Go to **Integrations**](https://console.cloud.google.com/monitoring/integrations)

   <br />

   If you use the search bar to find this page, then select the result whose subheading is
   **Monitoring**.
2. Click the **Kubernetes Engine** deployment-platform filter.
3. Locate the GKE Inference Gateway integration and click **View Details**.
4. Select the **Dashboards** tab.

<br />

## Troubleshooting

For information about troubleshooting metric ingestion problems, see
[Problems with collection from exporters](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/troubleshooting.md.txt#exporter-problems) in [Troubleshooting ingestion-side problems](https://docs.cloud.google.com/stackdriver/docs/managed-prometheus/troubleshooting.md.txt#ingest-problems).