---
title: "Grafana monitoring: let the problem ring first"
description: "Prometheus, Loki and Grafana collection, alerting and retention; what gets delivered, and why to start with few accurate alerts."
canonical: "https://localdeploy.dev/services/grafana"
source: "content/pages/services/grafana.md"
lang: "en"
altPath: "/zh/services/grafana"
---


## When you need it

With several services, a full disk, an expiring certificate or a restart loop is discovered when something stops working. Monitoring is not about drawing graphs; it is about hearing the problem before you notice.

## What gets delivered

- Collection: node_exporter for the host, cAdvisor for containers, application metrics where they already exist.
- Storage and retention: where Prometheus data lives, how long it is kept, and how much disk that needs.
- Logs, optional: Loki for container logs, queried in the same Grafana.
- Dashboards: host, containers, certificate expiry, backup results; the few you will actually read.
- Alerts: rules, thresholds, notification channels (email or webhook), plus silencing and grouping so one issue does not ring all night.
- A test: fill a disk or stop a service once, and confirm the alert actually fires.

## Not included

- Hosted monitoring. The data lives on your machine, and if the machine is down, so is the monitoring.
- 24/7 on-call or picking up alerts for you.
- Third-party notification availability.
- Instrumenting application code; existing metrics are wired up.

## Who it fits

Several services, or a previous outage that was only noticed the next morning. One service that restarts now and then does not need this yet.

## How to start

Email how many machines, which services, whether Prometheus already exists, and where alerts should go. Scope first, then a quote.

## Questions

### Is it heavy?

Prometheus and Grafana are light; metric cardinality and retention are what cost. Collection scope and retention are sized to your machine.

### Too many alerts?

Start with four: disk, certificates, service down, backup failed. Tune thresholds after a week. More rules is not better.

## Related

- [k3d / k3s](/services/k8s)
- [Traefik reverse proxy](/services/traefik)
- [Pricing and payment](/pricing)

## Structured data

```json
{
  "@context": "https://schema.org",
  "@type": "WebSite",
  "name": "localdeploy.dev",
  "url": "https://localdeploy.dev/",
  "description": "localdeploy.dev helps home labs and small teams assess the hardware they already own, then deploy a local AI assistant with Cloudflare, self-hosting and containers. Free basics, fixed-scope deployment packages, clear boundaries."
}
```
