Skip to main content

Documentation Index

Fetch the complete documentation index at: https://ray-preview.mintlify.app/llms.txt

Use this file to discover all available pages before exploring further.

​
Open the dashboard

ray dashboard cluster.yaml          # forward port 8265 (cluster launcher)
kubectl port-forward svc/<head>-svc 8265:8265   # KubeRay
Then open http://localhost:8265.

​
Install Prometheus and Grafana

The Ray repo ships Helm charts for both. On Kubernetes: 
helm repo add kuberay https://ray-project.github.io/kuberay-helm/
helm install kuberay-monitoring kuberay/kuberay-monitoring -n monitoring --create-namespace
On VMs, install your preferred Prometheus distribution and point its scrape config at the Ray service-discovery file: 
scrape_configs:
  - job_name: ray
    file_sd_configs:
      - files: [/tmp/ray/prom_metrics_service_discovery.json]

​
Import dashboards

Ray writes default Grafana dashboards to /tmp/ray/session_latest/metrics/grafana/dashboards/. Import them through the Grafana UI or sidecar provisioning.

​
Tail logs

ray exec cluster.yaml "tail -f /tmp/ray/session_latest/logs/raylet.out"
For aggregated logs, ship /tmp/ray/session_latest/logs/ to your log backend with fluentbit or similar.

​
Capture a flame graph

From the dashboard, find the worker on the Tasks or Actors tab and click “Stack Trace” to capture a py-spy snapshot. Or from CLI: 
ray exec cluster.yaml "py-spy record -d 30 -p <worker-pid> -o /tmp/profile.svg"

​
Next steps

Metrics

Per-metric reference.

State API

Programmatic access to cluster state.