Diagnostics

The Diagnostics page is the agent’s system-health surface: hardware utilization, temperatures, systemd unit state, and quick actions for restart and reboot.

System panel

CPU usage, core count, memory total / used / percent, disk total / used / percent, and the temperature sensors the kernel exposes (cpu_thermal, SoC, board-level). The page polls /api/system every five seconds.

Control-plane latency

Mission Control measures the round-trip time of its control link to the agent and shows it as controlRttMs on the drone card and the agent status card. It is the time for a request to reach the agent and a response to come back, so it reflects the path the GCS is actually using (LAN WebSocket, MQTT relay, or the cloud poll), independent of the flight controller. The agent answers a lightweight GET /api/ping echo for this measurement, and the GCS also derives a value from its regular status poll, so a number shows up without any extra configuration.

Reading	What it means
Tens of milliseconds	Direct LAN connection, healthy.
100-300 ms	MQTT relay or a busy network.
Seconds	Cloud command-queue poll, or a degraded link. Manual control will feel laggy. Prefer a LAN connection for hands-on flying.

A missing or steadily climbing value means the control path is unhealthy. Check the connection method (LAN versus cloud) and the network between the GCS and the agent.

Services panel

Every systemd unit in the agent’s set. Each row shows unit name, active / inactive, sub-state (running, dead, failed), and PID.

Per-service actions

Action	Effect
Restart	`systemctl restart` on the unit.
Stop	Stop without disabling.
Logs	Filter the Logs page to this service.

Reboot

A reboot button at the bottom of the page calls /api/v1/setup/reboot. The reboot is graceful: the agent flushes state, stops services, and schedules the reboot.

Profile gating

Shared route, visible on both profiles.

Backend endpoints

Method	Path	Purpose
GET	`/api/system`	CPU / memory / disk / temps
GET	`/api/services`	Unit list
GET	`/api/ping`	Lightweight echo for control-plane round-trip timing
POST	`/api/services/{name}/restart`	Restart a unit
POST	`/api/v1/setup/reboot`	Reboot the node

Troubleshooting

Service list reads `service inventory unavailable`

The agent could not reach its own dbus connection to systemd. Run systemctl status ados-supervisor over SSH. A typical cause is a permission misconfiguration on the unit file.

A service keeps restarting

Open the service’s logs from the row action. The most common causes are missing hardware (the service waits forever for a device), a config typo (the service crashes on parse), or an upstream dependency the service waits on. Restart with the dependency healthy.

​Diagnostics

​System panel

​Control-plane latency

​Services panel

​Per-service actions

​Reboot

​Profile gating

​Backend endpoints

​Troubleshooting

​Related