Setting up observability¶
make observe starts Prometheus, Grafana, and Grafana Image Renderer for the deployed fleet. Prometheus collects router and engine metrics and evaluates alert rules. Grafana serves the provisioned Narwhal Orchestrator dashboard.
Monitoring tasks¶
-
Starting and verifying monitoring
Configure the monitored deployment, start Prometheus and Grafana, and verify targets.
-
Accessing dashboards and isolating listeners
Reach the dashboard from a workstation and isolate a second monitoring stack.
-
Read each Narwhal Orchestrator panel and engine state.
-
GPU telemetry, alerts, and recovery
Feed GPU telemetry, inspect alerts, troubleshoot monitoring, and retain captures.
-
Monitoring a WSL2 development fleet
Connect, start, and check monitoring for a WSL2 development fleet.