Prometheus metrics¶

OpenWatch exposes your cost and resource figures as Prometheus metrics, so you can scrape them into your own monitoring stack and build dashboards or alerts alongside the rest of your infrastructure.

Endpoint¶

GET /prometheus/metrics

The endpoint returns metrics in the standard Prometheus text exposition format. It is authenticated with a per-organisation bearer token and is exempt from CSRF (it is a machine-to-machine endpoint that authenticates itself).

Authentication¶

Each organisation has its own Prometheus token. Pass it in the Authorization header:

Authorization: Bearer <your-token>

Requests without a valid token receive 401 Unauthorized. Because the token scopes the response to a single organisation, keep it secret and rotate it if it leaks.

One token per organisation

The metrics returned are those of the organisation that owns the token. If you run several organisations, scrape each with its own token (and its own Prometheus job or target label).

Scrape configuration¶

Add a job to your prometheus.yml:

scrape_configs:
  - job_name: openwatch
    scheme: https
    bearer_token: "<your-token>"
    metrics_path: /prometheus/metrics
    static_configs:
      - targets: ["openwatch.example.com"]

Replace the target with your OpenWatch host and the token with your organisation's token.

Metrics exposed¶

Metric	Type	Description
`openwatch_cost_compute_monthly`	gauge	Estimated monthly compute cost.
`openwatch_cost_storage_monthly`	gauge	Estimated monthly storage cost.
`openwatch_cost_network_monthly`	gauge	Estimated monthly network cost.
`openwatch_cost_total_monthly`	gauge	Estimated monthly total cost.
`openwatch_resource_count`	gauge	Number of resources, labelled by `type` (for example `servers`, `volumes`, `routers`).
`openwatch_rightsizing_savings_monthly`	gauge	Estimated monthly savings identified by rightsizing.
`openwatch_rightsizing_score`	gauge	Rightsizing efficiency score.
`openwatch_rightsizing_recommendations`	gauge	Number of open rightsizing recommendations.

Cost values follow the same figures shown on the cost dashboard. Resource counts mirror the resource explorer.

Freshness

Metrics reflect OpenWatch's cached data, which is refreshed in the background. They are not a live query against your cloud on every scrape, so a short scrape interval will not increase load on your cloud provider.