How it works
- An alert fires in Grafana
- Foggy receives the alert with its full context (name, labels, severity, duration)
- Foggy automatically starts an investigation using the alert as the starting point
- Results are delivered to Slack (if configured) and stored as a thread in the web interface
Setup
Go to Connectors → Add Connector → Grafana Alerts and fill in the form.| Field | Required | Description |
|---|---|---|
| Name | Yes | A label for this alert source (e.g. “Production Grafana Alerts”) |
| Host | Yes | Your Grafana instance URL (e.g. https://my-company.grafana.com) |
| Token | No | API token for accessing alerts |
Alert storm handling
Foggy handles alert storms intelligently using a queue-based approach:- Bundling — If an investigation is already running, incoming alerts are queued and investigated together in a single bundled pass when the queue drains. During a cascading failure with 20 alerts, you get one bundled investigation that connects the dots — not 20 separate ones.
- Dedup — The same alert firing repeatedly while queued counts as one entry (with a fire count), not separate investigations.
Pairing with automations
Use Grafana Alerts as a trigger for alert-triggered automations. Define filter rules to match specific alert labels so only relevant alerts trigger investigations.Foggy uses read-only access. Your data is never modified.