EdgeObservabilityAgent

EdgeObservabilityAgent is a .NET 9 solution for collecting edge-node telemetry, buffering it locally during outages, and forwarding it to a lightweight HQ collector when connectivity is available.

Solution layout

• EdgeObservabilityAgent/
  Worker service that:

  • collects CPU, memory, and Docker container metrics every 5 seconds
  • stores metrics in a local SQLite buffer (edge_buffer.db)
  • attempts periodic resync to HQ (http://localhost:5100)
  • renders a live terminal dashboard (interactive terminals only)

• HqCollector/
  Minimal ASP.NET Core API that:

  • exposes GET /health for connectivity checks
  • accepts telemetry via POST /ingest
  • exposes latest per-node snapshots via GET /metrics

Project files

• EdgeObservabilityAgent.sln — solution entry containing both projects
• EdgeObservabilityAgent/Program.cs — dependency registration and host startup
• EdgeObservabilityAgent/Worker.cs — metric collection loop and dashboard update flow
• EdgeObservabilityAgent/Services/BufferRepository.cs — SQLite buffering and sync state storage
• EdgeObservabilityAgent/Services/ResyncService.cs — connectivity checks and buffered payload resync
• EdgeObservabilityAgent/Services/SystemMetricsService.cs — CPU/memory/Docker metric collection
• EdgeObservabilityAgent/Services/DashboardService.cs — live terminal dashboard rendering
• EdgeObservabilityAgent/Models/MetricPayload.cs — payload/domain models
• EdgeObservabilityAgent/Protos/metrics.proto — protobuf schema for metrics
• HqCollector/Program.cs — minimal API with /health, /ingest, and /metrics

Prerequisites

• .NET SDK 9.0+
• Docker (optional, for container telemetry)

Build

From repository root:

dotnet restore EdgeObservabilityAgent.sln
dotnet build EdgeObservabilityAgent.sln

Run locally

Open two terminals.

1) Start HQ collector

dotnet run --project HqCollector/HqCollector.csproj

Collector listens on http://localhost:5100.

2) Start edge agent

dotnet run --project EdgeObservabilityAgent/EdgeObservabilityAgent.csproj

The agent will:

• buffer payloads in SQLite
• detect HQ availability via /health
• flush buffered rows to /ingest when connected

API quick reference (HQ collector)

• GET /health → liveness check
• POST /ingest → accepts JSON metric payload
• GET /metrics → returns latest payload and retained count per node

Data model

Primary payload fields include:

• nodeId
• timestamp
• cpuUsage
• memoryUsageMb
• memoryTotalMb
• containers[] (id, name, status, image)

See EdgeObservabilityAgent/Protos/metrics.proto for schema details.

Notes

• Resync polling runs every 2 seconds in batches of up to 50 records.
• Dashboard output is automatically suppressed in non-interactive environments.