Prometheus metrics
Scrape GET /metrics for these gauges. The endpoint is hand-rolled Prometheus
text format with no dependency and no cache. A value with no current reading (a failed
test, or no tests yet) renders as NaN so Prometheus doesn’t carry a stale value;
baudflow_health is omitted entirely when there’s no verdict (calibrating, off
mode, failed, or no latest).
The bandwidth gauges reflect the latest Ookla speed test; a ping-only measurement doesn’t populate them.
| Metric | Description |
|---|---|
baudflow_download_mbps |
Latest download speed (Mbps). |
baudflow_upload_mbps |
Latest upload speed (Mbps). |
baudflow_ping_latency_ms |
Latest latency (ms). |
baudflow_ping_jitter_ms |
Latest jitter (ms). |
baudflow_packet_loss |
Latest packet loss (raw Ookla value). |
baudflow_health |
1 if the latest test is healthy, 0 if unhealthy. Omitted when there’s no verdict. |
baudflow_measurements_total |
Total retained measurements. |
baudflow_uptime_percentage |
Percentage of tests over the window that were healthy. |
Scraping
scrape_configs:
- job_name: baudflow
static_configs:
- targets: ['baudflow:4000']Point a Grafana panel or an Alertmanager rule at it and your homelab observability stack picks baudflow up like any other exporter.