Sign in Subscribe
how-to

How to Connect Google Cloud Managed Prometheus to Grafana Using Docker Compose (2025 Setup Guide)

Huzaif Shah

• 3 min read

Table of Contents

Introduction

Google Cloud Managed Prometheus (GMP) lets you store and query Prometheus metrics in Google Cloud Monitoring without running your own Prometheus servers. If you want a familiar Prometheus-style experience in Grafana, you can run the Prometheus query frontend container locally and point Grafana’s Prometheus data source at it. The query frontend authenticates to Google Cloud using a service account and translates PromQL queries to Google Cloud Monitoring API calls.

This guide shows a minimal, production-friendly setup using Docker Compose. You’ll create a least-privilege service account, run the managed Prometheus query frontend, and connect Grafana to it. The end result: you can explore your Google Cloud metrics with PromQL in Grafana, without operating Prometheus storage yourself.

Prerequisites

  • Docker and Docker Compose installed
  • Google Cloud project ID where your GMP metrics live
  • gcloud CLI authenticated with permissions to create service accounts and IAM bindings

Create a service account and key (least privilege)

Replace PROJECT_ID with your Google Cloud project ID.

gcloud config set project PROJECT_ID

# Create a dedicated service account for Grafana → GMP reads
gcloud iam service-accounts create grafana-gmp-reader \
  --display-name="Grafana GMP Reader"

# Grant read-only access to metrics
gcloud projects add-iam-policy-binding PROJECT_ID \
  --member="serviceAccount:grafana-gmp-reader@PROJECT_ID.iam.gserviceaccount.com" \
  --role="roles/monitoring.viewer"

# Enable the Monitoring API if not already enabled
gcloud services enable monitoring.googleapis.com

# Create a JSON key (store securely!)
gcloud iam service-accounts keys create service-account-key.json \
  --iam-account=grafana-gmp-reader@PROJECT_ID.iam.gserviceaccount.com

Security notes:

  • Prefer avoiding long-lived keys in production. If you’re running on GKE or GCE, use Workload Identity instead of keys. For hybrid/self-hosted, consider Workload Identity Federation.
  • Scope permissions to only the projects you need. roles/monitoring.viewer is typically sufficient for read-only querying.

Docker Compose for Grafana and the Prometheus query frontend

Create a docker-compose.yaml with the following content. Replace YOUR_GCP_PROJECT_ID with your project and point the volume to your downloaded key file.

version: '3.8'

services:
  grafana:
    image: grafana/grafana:latest
    container_name: grafana
    ports:
      - "3000:3000"
    environment:
      - GF_SECURITY_ADMIN_USER=admin
      - GF_SECURITY_ADMIN_PASSWORD=change-me
    volumes:
      - grafana-storage:/var/lib/grafana
    restart: unless-stopped
  
  prometheus:
    image: gke.gcr.io/prometheus-engine/frontend:v0.15.3-gke.0
    container_name: gmp-query-frontend
    ports:
      - "9090:9090"
    environment:
      - GMP_PROJECT=YOUR_GCP_PROJECT_ID
      - GOOGLE_APPLICATION_CREDENTIALS=/etc/gcp/service-account-key.json
    volumes:
      - ./service-account-key.json:/etc/gcp/service-account-key.json:ro
    command:
      - "--web.listen-address=:9090"
      - "--query.project-id=YOUR_GCP_PROJECT_ID"
    restart: unless-stopped

volumes:
  grafana-storage:

Bring the stack up:

docker compose up -d

Configure Grafana to use the Prometheus query frontend

  1. Open Grafana at <http://localhost:3000> (default user admin, password you set).
  2. Navigate to Connections → Data sources → Add data source → Prometheus.
  3. Set URL to <http://prometheus:9090> (the Docker Compose service name and port).
  4. Save & test.

Validate with a simple query

In Grafana Explore, select the Prometheus data source and try queries such as:

up

If you have GMP scraping configured in your project, you can also try common container metrics like:

container_cpu_usage_seconds_total

If queries return no data, double-check:

  • The project ID in Docker Compose flags and GMP_PROJECT
  • The service account role (roles/monitoring.viewer)
  • That Google Managed Prometheus is ingesting metrics in your project

Conclusion

With a lightweight Docker Compose setup, you can explore Google Cloud Managed Prometheus metrics in Grafana using standard PromQL—no Prometheus storage to manage. Keep credentials safe, prefer identity-based access over keys where possible, and consider provisioning for repeatable environments. From here, import or build dashboards targeting <http://prometheus:9090> and iterate on your observability views.

Recent Posts