Deep Telemetry for Unifi with Unpoller and Prometheus

Introduction

While the Ubiquiti Unifi interface provides a great overview of your network, it lacks the deep, long-term metric retention and granular visualization capabilities of a dedicated monitoring stack. To bridge this gap, I integrated Unpoller (formerly UniFi-Poller) into my Home Lab’s Prometheus and Grafana ecosystem.

The Architecture

Unpoller acts as a data bridge. It polls the Unifi Controller’s API and exports the data in a format that Prometheus can easily ingest.

1. Secure Credential Management

Since my Home Lab configuration is managed via GitOps, security is paramount. I used Bitnami SealedSecrets to store my Unifi read-only credentials. This allows me to commit the encrypted secret to my public repository safely.

2. Deployment and Service Discovery

I deployed Unpoller as a lightweight container in my monitoring namespace. To automate the metrics collection, I used a ServiceMonitor resource:

apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: unpoller
  labels:
    release: grafana-prometheus-monitoring
spec:
  selector:
    matchLabels:
      app: unpoller
  endpoints:
  - port: metrics
    interval: 30s

By adding the release label, the Prometheus Operator automatically discovers the new target and begins scraping.

The Results

With Unpoller running, I now have access to:

• Per-client bandwidth usage over months, not just hours.
• Detailed USG telemetry, including CPU load, memory, and WAN latency.
• Port-level stats for all my Unifi switches.

All of this data is visualized in Grafana (using the excellent Dashboard ID 11310), giving me a single pane of glass for both my Kubernetes cluster and my physical network infrastructure.

Conclusion

Unpoller is a must-have for any Unifi user running a Prometheus stack. It turns your network into a measurable, observable component of your lab, allowing for better troubleshooting and capacity planning.