Prometheus metrics

Lookout can expose the latest collected values as Prometheus text, so you can scrape them into Prometheus, Grafana Agent, VictoriaMetrics, or anything that speaks the Prometheus exposition format. It is off by default.

This serves only the current in-memory snapshot — it does not add storage, retention, dashboards, or a query layer. Lookout is an alerting agent first; this endpoint is a read-only export for systems that already collect metrics.

Enable it

metrics:
  enabled: true
  listen: "127.0.0.1:9100"

Then restart the service:

sudo systemctl restart lookout

Lookout serves GET /metrics on the listen address (it also answers HEAD).

Field	Default	Description
enabled	false	Turn the endpoint on
listen	127.0.0.1:9100	host:port to serve /metrics on

listen defaults to loopback. Keep it on 127.0.0.1 or a private interface unless it sits behind a firewall or reverse proxy — the endpoint is unauthenticated.

What it exposes

Every reading is a Prometheus gauge carrying a host label. A scrape looks like:

# TYPE memory_used_percent gauge
memory_used_percent{host="web-1"} 73.42 1718790000000
# TYPE cpu_used_percent gauge
cpu_used_percent{host="web-1"} 18.6 1718790000000
# TYPE swap_used_percent gauge
swap_used_percent{host="web-1"} 4.1 1718790000000
# TYPE load_1_per_core gauge
load_1_per_core{host="web-1"} 0.62 1718790000000
# TYPE disk_used_percent gauge
disk_used_percent{host="web-1",mount="/"} 51.0 1718790000000

Disk metrics carry a mount label so each filesystem is its own series.

Service and presence checks export a 0 / 1 health gauge (1 means unhealthy or missing), labelled by name:

# TYPE systemd_unhealthy gauge
systemd_unhealthy{host="web-1",name="nginx"} 0
# TYPE http_unhealthy gauge
http_unhealthy{host="web-1",name="app"} 0
# TYPE tcp_unhealthy gauge
tcp_unhealthy{host="web-1",name="redis"} 0
# TYPE process_missing gauge
process_missing{host="web-1",name="nginx"} 0

The values mirror what the alert rules evaluate, so you can build dashboards and your own Prometheus alerting rules on the same numbers Lookout uses.

Scraping it

Point your scraper at the listen address. A minimal Prometheus job:

scrape_configs:
  - job_name: lookout
    static_configs:
      - targets: ["10.0.0.5:9100"]

If you keep listen on 127.0.0.1, scrape from the same host or reach it over a private network / tunnel. Lookout does not add authentication to the endpoint, so don't expose it directly to the public internet.