Skip to content

Observability

Atlas exposes a Prometheus /metrics endpoint and structured logs, plus a terminal inspection tool for watching a fleet live — the stand-in for the web console (which is its own later milestone).

The gateway serves Prometheus /metrics, including a queue-depth / utilization signal usable for external autoscaling (e.g. an ASG scaling policy). The endpoint requires an admin-scoped key, so configure your scraper to send it — mint one with atlas keys create --admin and pass it as a Authorization: Bearer <key> (or x-api-key) header in the Prometheus scrape config. Then build dashboards as usual.

Terminal window
atlas status # a snapshot of the fleet
atlas top # a live view — run over SSH on the gateway box

atlas top is the “see what the fleet is doing” view from the terminal: workers, deployed models, in-flight requests, and load.

Under load Atlas sheds with retryable codes rather than timing out — 429 rate_limit_error (momentarily full) and 529 overloaded_error (no live replica right now), both carrying Retry-After. Watching these alongside queue depth tells you when to add capacity. See API compatibility and Scaling & operations.