Asoba Ona Documentation
Home Insights Contact Us

Deploy the Edge Stack

This guide deploys the full Ona Platform stack on Ubuntu 22.04 — real-time monitoring, LSTM/TCN forecasting, JEPA anomaly detection, and automated maintenance workflows. No cloud account required.

Hardware Requirements

Component Minimum Recommended Notes
CPU 4 cores 8+ cores Training and inference are CPU-bound without a GPU
RAM 8 GB 16 GB 8 GB supports inference only; 16 GB supports full stack + training
Disk 20 GB free 50+ GB Docker images (~5 GB), PostgreSQL data, and ML artifacts grow over time
GPU None NVIDIA 4+ GB VRAM Optional — enables GPU-accelerated LSTM training and TensorFlow inference
Network LAN access LAN + WAN WAN needed for remote SSH and Huawei FusionSolar API

RAM tiers

RAM Forecasting Inference Site-Specific Training Full Stack + Training
8 GB ✅ (stop non-essential services first) ❌ ❌
12 GB ✅ ✅ (stop heavy services during training) ❌
16 GB ✅ ✅ ✅
24 GB+ ✅ ✅ ✅ (with headroom)

Preflight Check

Run these before starting to verify the system meets requirements:

nproc                                                    # CPU cores (need ≥ 4)
free -m | awk '/^Mem:/{print $2}'                        # RAM in MB (need ≥ 8000)
df -h / | awk 'NR==2{print $4}'                          # Free disk (need ≥ 20G)
nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null \
  || echo "No NVIDIA GPU (CPU-only mode)"
docker --version
python3 --version

Minimum: 4 CPU cores, 8 GB RAM, 20 GB free disk, Docker installed, Python 3.8+.

Prerequisites

Docker Engine

sudo apt update && sudo apt install -y docker.io docker-compose-plugin
sudo systemctl enable docker && sudo systemctl start docker
sudo usermod -aG docker $USER
# Log out and back in for the group change to take effect

Python dependencies

sudo apt install -y python3 python3-pip
pip3 install psycopg bcrypt

Clone the repository

git clone https://github.com/AsobaCloud/platform.git ~/platform
cp ~/platform/docker-compose.local.yml ~/docker-compose.local.yml

The compose file must live in the home directory (~). Running it from ~/platform/ creates a conflicting Docker project name.

API Keys

Key Required For How to Obtain
VISUAL_CROSSING_API_KEY Weather data for forecasting Visual Crossing — free tier available
HUAWEI_USERNAME Huawei FusionSolar inverter data Huawei FusionSolar admin account
HUAWEI_SYSTEM_CODE Huawei FusionSolar API auth Huawei Northbound API key

Add to ~/.bashrc for persistence across reboots:

echo 'export VISUAL_CROSSING_API_KEY="<your-key>"' >> ~/.bashrc
echo 'export HUAWEI_USERNAME="<your-username>"' >> ~/.bashrc
echo 'export HUAWEI_SYSTEM_CODE="<your-code>"' >> ~/.bashrc
source ~/.bashrc

The platform starts without Huawei credentials if you are not collecting from Huawei inverters. The Visual Crossing key is required for weather-dependent forecasting.

Installation

bash ~/platform/scripts/local/install.sh

The installer is idempotent — safe to re-run. It performs these steps:

Step Action
1 Preflight checks (Docker, Python, psql)
2 Start PostgreSQL and Redis
3 Initialize databases (ona_telemetry, ona_auth, ona_control, pvinsight)
4 Seed auth (admin/admin) and control-plane baseline data
5 Start all microservices and copy shared utilities into containers
6 Deploy Nginx edge proxy and web UI
7 Verify key API endpoints (auth, customers, telemetry)

Post-Install Verification

# Auth — should return a JWT
curl -s -X POST http://localhost/api/auth/login \
  -H 'Content-Type: application/json' \
  -d '{"username":"admin","password":"admin"}' | python3 -m json.tool

# Forecasting API health
curl -s http://localhost:8002/health | python3 -m json.tool

# Weather cache — trigger a forecast fetch
curl -s -X POST http://localhost:8004/run \
  -H 'Content-Type: application/json' \
  -d '{"mode": "forecast"}' | python3 -m json.tool

Accessing the Platform

Local (on the device)

URL Description
http://localhost/ Platform web UI
http://localhost/api/auth/login Auth endpoint
http://localhost:8001/ terminalApi (OODA, nowcast, assets)
http://localhost:8002/health forecastingApi health
http://localhost:8005/ anomalyModelService

Default credentials: admin / admin. Change these before exposing to a network.

Remote access via SSH

# Enable SSH on the edge device
sudo apt install -y openssh-server
sudo systemctl enable ssh && sudo systemctl start ssh

# Tunnel the web UI from a remote machine
ssh -L 8080:localhost:80 <username>@<device-ip>
# Then open http://localhost:8080 in your browser

# Tunnel multiple ports at once
ssh -L 8080:localhost:80 -L 8001:localhost:8001 -L 8002:localhost:8002 <username>@<device-ip>

Remote access via LAN

Find the device IP:

hostname -I   # on the edge device

Then open http://<device-ip>/ from any machine on the same LAN.

WAN access via jump host

# ~/.ssh/config on your remote machine
Host edge-device
    HostName 127.0.0.1
    Port 22
    User <edge-user>
    ProxyJump <public-jump-host>
    ServerAliveInterval 30
    ServerAliveCountMax 3

For production WAN access, consider WireGuard VPN or reverse SSH tunnel rather than exposing ports directly.

Next Steps