<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

NVIDIA Triton. This document shows you how to do the following:

- Set up Triton 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

[Triton documentation](https://docs.nvidia.com/deeplearning/triton-inference-server/user-guide/docs/user_guide/metrics.html)

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 NVIDIA Triton, see [Triton](https://developer.nvidia.com/triton-inference-server).

For information about setting up Triton on Google Kubernetes Engine,
see the GKE [guide for Triton](https://docs.cloud.google.com/kubernetes-engine/docs/tutorials/online-ml-inference.md.txt).

## Prerequisites

To collect metrics from

Triton

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 />

Triton exposes Prometheus-format metrics automatically; you do not have to install it separately. To verify that Triton is emitting metrics on the expected endpoints, do the following:

<br />

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

   ```
   kubectl -n NAMESPACE_NAME port-forward POD_NAME 8002:8002
   ```
2. Access the endpoint `localhost:8002/metrics` by using the browser or the `curl` utility in another terminal session.

## Define a PodMonitoring resource

For target discovery, the Managed Service for Prometheus Operator
requires a PodMonitoring resource that corresponds to
Triton in the same namespace.

You can use the following PodMonitoring 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: PodMonitoring
metadata:
  name: triton
  labels:
    app.kubernetes.io/name: triton
    app.kubernetes.io/part-of: google-cloud-managed-prometheus
spec:
  endpoints:
  - port: 8002
    scheme: http
    interval: 30s
    path: /metrics
  selector:
    matchLabels:
      app: triton
```
Ensure that the values of the `port` and `matchLabels` fields match those of the Triton 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
Triton. 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:  

   ```
   up{job="triton", cluster="CLUSTER_NAME", namespace="NAMESPACE_NAME"}
   ```

<br />

## View dashboards

The Cloud Monitoring integration includes

the **Triton 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, **Triton 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 NVIDIA Triton 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).