Fleet compute node for the A.R.C.H.I.E. platform. Self-registers with hub, reports hardware, heartbeats every 30s.
docker run -d --name archie-node --restart unless-stopped \
--gpus all \
-e HUB_URL=http://your-hub:3000 \
-e TOKEN=your-registration-token \
-p 3001:3001 \
-v archie-node-data:/data \
ghcr.io/kytrankatarn/archie-fleet-node:latestWithout GPU: Remove --gpus all.
- Container starts, detects hardware (CPU, RAM, GPU)
- Registers with hub using TOKEN for instant approval
- Heartbeats every 30s with system metrics
- Local API on port 3001 for health checks
- Connects to local Ollama (if installed) for LLM dispatch
| Variable | Required | Default | Description |
|---|---|---|---|
HUB_URL |
Yes | - | Hub URL (e.g., http://192.168.1.200:3000) |
TOKEN |
No | - | Registration token for auto-approval |
NODE_NAME |
No | hostname | Display name in hub |
OLLAMA_HOST |
No | http://host.docker.internal:11434 |
Ollama API URL |
HEARTBEAT_INTERVAL |
No | 30 |
Seconds between heartbeats |
NODE_API_PORT |
No | 3001 |
Local API port |
- Docker Desktop (Windows/Mac) or Docker Engine (Linux)
- Optional: NVIDIA GPU + drivers for GPU workloads
- Optional: Ollama installed separately for LLM inference
Apache 2.0 — Kytran Empowerment Inc.