Grafana Overview Dashboard#
Added in v0.5. Ships as deploy/grafana/infraguard.json in the repo.
One import gets you the standard set of panels:
- Request rate (per minute)
- Blocked rate (per minute)
- Active beacons (5-minute window)
- Upstream p95 latency
- Requests by filter_result over time
- Per-filter block rate
- Upstream latency p50 / p95 / p99
- Circuit-breaker state changes
- Top 20 blocked IPs in the last 15 minutes (table)
- Per-domain burn score
Import#
curl -X POST -H "Content-Type: application/json" \
-d @deploy/grafana/infraguard.json \
-u admin:$GRAFANA_PASSWORD \
http://your-grafana:3000/api/dashboards/dbOr paste the JSON into Grafana’s Dashboards > Import dialog.
Metrics referenced#
The panels expect these Prometheus series, all emitted by
infraguard/ui/api/metrics.py or by the prom_custom,
beacon_correlation, and burn-scoring paths:
infraguard_requests_total{filter_result, filter_name, domain}infraguard_filter_hits_total{filter_name, action}infraguard_upstream_duration_ms_bucket{le, upstream}infraguard_breaker_transitions_total{upstream, state}infraguard_active_beaconsinfraguard_burn_score{domain}
Missing data on some panels?#
If infraguard_active_beacons or infraguard_burn_score show No data
after import, the corresponding paths (state/beacons.py beacon
correlation, or the burn scorer) may not be enabled in your config.
Confirm the series exists by hitting /metrics directly on the
dashboard container.
The other panels start populating as soon as the proxy sees traffic; allow ~30 seconds for the histogram buckets to fill on a fresh scrape.
Custom panels#
Use the prom_custom plugin to emit operator-defined labels (country,
campaign, target group) and add panels that break request rate down by
those dimensions.
plugin_settings:
prom_custom:
options:
counters:
- name: infraguard_requests_by_country_total
labels: ["country"]
from: "metadata.geo.country"See prometheus-metrics for the base metric inventory.