Skip to main content

Prometheus export

GET https://api.wakora.io/v1/prometheus answers in the Prometheus exposition format, so a Prometheus, a Grafana Agent or anything else that scrapes can pull the fleet into the dashboards you already have. It needs a token with read scope, on the Pro plan and above.

What one scrape returns​

Host meters, one series per host, labelled with host (the display name) and server (the UUID):

  • wakora_host_up (1 or 0), wakora_host_cpu_used_pct, wakora_host_mem_used_pct, wakora_host_disk_used_pct (the fullest disk), wakora_host_last_seen_age_seconds;
  • wakora_host_info with os, platform and agent_version labels, value 1, the usual info-metric pattern for joins.

Incidents: wakora_incidents{state,severity} counts open and acknowledged incidents per severity, wakora_incidents_resolved_24h counts the last day's resolutions.

Usage: wakora_usage{axis} with the live value of every billing axis (hosts, sites, APM sites, devices, container groups, RUM views, log volume, flow volume, status pages).

Raw series​

Add server=<uuid> and names=<comma-separated exact names> to scrape raw agent metrics of one host, up to 40 names per call:

/v1/prometheus?server=b62a...&names=host.load1_per_core,svc.mysql.running

They come back as wakora_metric{name="host.load1_per_core",server="b62a..."} with the latest value. Metric names are in the metrics reference; /v1/metrics/names?server= lists what a host actually produces.

Scrape configuration​

scrape_configs:
  - job_name: wakora
    scheme: https
    metrics_path: /v1/prometheus
    static_configs:
      - targets: ["api.wakora.io"]
    authorization:
      type: Bearer
      credentials: wka_...
    scrape_interval: 60s

Sixty seconds matches the agents' metric cadence; scraping faster returns the same values more often. The endpoint is subject to the API rate limit, so a fleet of scrapers pointing at one workspace should share one job.

Notes​

  • The endpoint is a snapshot of the latest values, not a history export. Long ranges live in the metrics API with rollups.
  • A host that has not reported for a while keeps wakora_host_last_seen_age_seconds growing and wakora_host_up at 0; it disappears from the export when it is removed from the workspace.