Skip to main content
GET /metrics serves app, database, node and deploy metrics in the Prometheus text format. It is never public: every request needs a read token, and a token limited to some apps only sees those apps.

Set up

This mints a read-only token named prometheus-scrape and prints a scrape config to paste into prometheus.yml. The token is shown once. Run levelrail-cli tokens scrape --rotate to replace it; the old token is revoked after the new one exists.

Turn it off

The observability.metrics_scrape_enabled runtime setting (default on) switches the endpoint without touching tokens. When off, /metrics answers 404. It applies live, from the dashboard, API or CLI.

What is exported

  • Per app, database and node: CPU, memory, network and disk metrics, labelled app, database or node.
  • Platform gauges: app count, deploy queue depth, running deploys, last and mean deploy duration, deploys finished by status, online agents, reconcile lag.
  • Export destination health, labelled destination.
Labels are limited to those names, so the series count tracks what you run and never grows with request paths or user input.
Last modified on November 17, 2026