I thought connecting Grafana to GCP Managed Prometheus would take 10 minutes.
It didn’t.
With a normal Prometheus setup, you just point Grafana to an endpoint and start querying metrics. But GCP Managed Prometheus works differently, and that’s where things get confusing.
Grafana was connected, but no metrics were showing — which made debugging frustrating.
Instead of a direct connection, you’re dealing with Google Cloud APIs, IAM permissions, and authentication layers. Small misconfigurations can break everything.
After digging into it, I realized most issues come down to authentication and endpoint configuration.
To simplify things, I set up Grafana using Docker Compose and worked through it step by step.
This approach helped me:
Run Grafana in a controlled environment
Fix authentication issues step by step
Confirm whether metrics were actually being returned
Once everything was aligned, the setup worked as expected.
If you’re working with GKE, Kubernetes monitoring, or managed observability on GCP, this is something you’ll likely run into.
I documented the full setup with a working example and exact configuration here:
https://www.kubeblogs.com/how-to-connect-google-cloud-managed-prometheus-to-grafana-using-docker-compose-2025-setup-guide/
What part took you the most time — authentication, configuration, or getting metrics to show up?