Coordinator Setup

The labgrid coordinator provides centralized management of hardware resources across multiple exporter hosts. The adi-labgrid-plugins project includes a Docker-based deployment of the coordinator, a REST/WebSocket API bridge, and a web dashboard.

Architecture

        graph LR
    subgraph Docker ["Docker Compose"]
        C[Coordinator<br/>gRPC :20408]
        A[API Server<br/>FastAPI :8000]
        W[Web Dashboard<br/>nginx :3000]
    end

    E1[Exporter Host 1] -->|gRPC ExporterStream| C
    E2[Exporter Host 2] -->|gRPC ExporterStream| C
    A -->|gRPC ClientStream| C
    W -->|/api proxy| A
    B[Browser] -->|HTTP + WebSocket| W
Coordinator web dashboard

The stack consists of three Docker services:

  • Coordinator: The labgrid coordinator gRPC server. Tracks places, resources, exporters, and reservations. Persists state to a Docker volume.

  • API: A FastAPI server that connects to the coordinator as a gRPC client, caches all state in memory, and exposes a REST + WebSocket API.

  • Web: An nginx server hosting the React dashboard and proxying /api requests to the API service.

Exporters run on physical hosts (not in Docker) and connect to the coordinator over the network.

Prerequisites

  • Docker and Docker Compose

  • Network connectivity between exporter hosts and the coordinator

Quick Start

cd coordinator
docker compose up -d

This starts all three services:

  • Coordinator at localhost:20408

  • API at localhost:8000

  • Web dashboard at http://localhost:3000

Configuration

Copy .env.example to .env and adjust ports if needed:

# coordinator/.env
LG_COORDINATOR_PORT=20408
API_PORT=8000
WEB_PORT=3000

Connecting Exporters

From any host with labgrid and the ADI plugins installed:

labgrid-exporter -c <coordinator-host>:20408 -n my-exporter resources.yaml

Or set the environment variable:

export LG_COORDINATOR=<coordinator-host>:20408
labgrid-exporter -n my-exporter resources.yaml

See Exporter Setup for resource YAML configuration details.

Testing Without Hardware

A mock exporter is included for development and testing:

cd coordinator
docker compose up -d  # starts coordinator, api, web

# Run the mock exporter (connects to coordinator)
cd mock-exporter
python mock_exporter.py -c localhost:20408

The mock exporter registers fake VCU118 and Raspberry Pi resources, allowing you to test the full dashboard workflow without any hardware.

API Reference

The API server exposes the following endpoints:

Health

  • GET /api/health - Coordinator connectivity status

Places

  • GET /api/places - List all places

  • GET /api/places/{name} - Get place details

  • POST /api/places - Create a place (body: {"name": "..."})

  • DELETE /api/places/{name} - Delete a place

  • POST /api/places/{name}/acquire - Acquire a place

  • POST /api/places/{name}/release - Release a place

  • PUT /api/places/{name}/tags - Set tags (body: {"tags": {...}})

  • PUT /api/places/{name}/comment - Set comment

  • POST /api/places/{name}/matches - Add resource match

  • DELETE /api/places/{name}/matches - Remove resource match

Resources

  • GET /api/resources - List all resources (filters: ?exporter=, ?cls=, ?avail=)

  • GET /api/exporters - List exporters with grouped resources

Reservations

  • GET /api/reservations - List all reservations

  • POST /api/reservations - Create reservation

  • DELETE /api/reservations/{token} - Cancel reservation

  • POST /api/reservations/{token}/poll - Poll reservation status

WebSocket

  • WS /api/ws - Real-time updates. On connect, receives full state snapshot. Subsequent messages are incremental updates:

    {"type": "place_update", "data": {...}}
{"type": "resource_delete", "data": {"exporter": "...", "group": "...", "name": "..."}}