Google Managed Prometheus

Configure Robusta to work with Google Cloud’s Managed Prometheus service.

Prerequisites

Before setting up Robusta, make sure you have:

  • Google Managed Prometheus enabled

  • A Prometheus Frontend endpoint accessible from your cluster (If you don’t already have one, you can create it following the instructions here).

Quick Start

Add the following configuration to your generated_values.yaml file:

globalConfig:
    # Set this to the URL of your Prometheus Frontend endpoint, it may change depending on the namespace you deployed it to.
    prometheus_url: http://frontend.default.svc.cluster.local:9090
    check_prometheus_flags: false  # Required for Google Managed Prometheus

Then update Robusta.

Service Locations

Google Managed Prometheus typically deploys services in these locations:

Prometheus Frontend:

  • Namespace: default (or where you deployed it)

  • Service: frontend

  • Port: 9090

AlertManager:

  • Namespace: gmp-system

  • Service: alertmanager

  • Port: 9093

Verify your exact service names:

# Check frontend service
kubectl get svc -n default | grep frontend

# Check AlertManager service
kubectl get svc -n gmp-system | grep alertmanager

Configuration Notes

Warning

Google Managed Prometheus does not support the Prometheus flags API. Always set check_prometheus_flags: false.

  • The Prometheus Frontend must be deployed separately (it's not included by default)

  • Ensure all required exporters are configured for full functionality

  • Google Managed Prometheus uses a different architecture than standard Prometheus

Verification

After configuration, verify the integration:

  1. Check connectivity:

    kubectl run test-curl --image=curlimages/curl --rm -it -- \
    curl -v http://frontend.default.svc.cluster.local:9090/-/healthy

  2. Test with a demo alert:

    kubectl apply -f https://raw.githubusercontent.com/robusta-dev/kubernetes-demos/main/oomkill/oomkill_job.yaml

    You should receive an alert with metrics attached.

  3. Check Robusta logs:

    kubectl logs -n robusta deployment/robusta-runner | grep -i prometheus

Next Steps