🌹 CRose (China Rose) - All-in-One Lightweight Data Engine

Chinese Readme | AI Training Framework | Edge Management Framework

Connect Datas, Deliver Values

CRose is an integrated data platform designed for manufacturing and modern agriculture. It encapsulates the full stack capabilities from underlying protocol collection (Modbus/MQTT) to upper-level statistical analysis and UI visualization.

🌟 Why Choose CRose? Simple yet Powerful!

🚀 Out-of-the-box full‑stack platform

• A single docker-compose up launches the complete stack: Odoo, Node‑RED, IoTDB, Redis, GMqtt, etc.
• No manual selection, integration, or tuning. Industrial‑grade IoT data collection and management from day one.

📦 Scenario‑based flow template library

• Built‑in templates for 20+ common acquisition scenarios: Modbus RTU/TCP, OPC UA, MQTT, S7, …
• Covers typical equipment such as machine tools, injection molding machines, heat treatment furnaces, environmental sensors—no need to handcraft Node‑RED flows from scratch.

🧠 Natural language → auto‑generated acquisition flows

• Describe your need in everyday language (e.g., “read PLC D100 every 10s and alert if > 80”).
• The system matches templates, fills parameters, and produces executable Node‑RED flows—lowering the low‑code barrier dramatically.

📊 End‑to‑end observability of data acquisition

• Collection health board: see whether each datapoint is being collected and validation results.
• Throughput and storage statistics: total records, messages/sec, time‑series storage footprint.
• Resource monitoring: CPU/Memory/Network trends on edge nodes and servers to flag bottlenecks early.

🌐 Large‑scale edge fleet management

• Batch deployment, flow updates, version rollback, configuration drift detection.
• Designed for hundreds of Raspberry Pis/industrial PCs with fully remote ops.

✅ Native data quality governance

• Schema checks at ingestion (units, ranges, non‑null, etc.), flagging anomalies.
• Data quality reports (missing rate, latency distribution, duplication rate) to underpin trustworthy AI analytics.

🚀 Quick Start

Note: Versions prior to 1.0 are preview versions and are not recommended for production use.

Deployment

git clone https://github.com/feitasIoT/Crose.git
cd Crose
docker-compose up -d --build
# Start AI services
docker compose -f docker-compose-ai.yml up -d

Optional: configure a registry prefix for images in docker-compose.yml.

# Example: pull from an internal registry or mirror
export REGISTRY=registry.example.com/
docker compose up -d --build

If REGISTRY is empty, Compose uses the original image addresses.

Offline Image Import

If the deployment environment cannot pull images directly, you can first download the image packages (*.tar) from a networked machine and then import them into the target Docker host.

Single image package:

docker load -i your-image.tar

Import all image packages in the current directory:

for f in ./*.tar; do
  docker load -i "$f"
done

After the import completes, run:

docker compose up -d

You will find many containers started. The core stack comes from docker-compose.yml, and the AI stack comes from docker-compose-ai.yml:

• nginx
• frps
• nexus
• web
• db
• gmqtt
• iotdb
• redis
• nodered-prod
• nodered-staging
• ai
• ai-train
• vllm

Although many containers are started, you can complete all operations in the Crose Web interface without any concerns.

Getting Started

• Access via browser (Chrome, Edge, etc.): http://ip:18069 Username: admin, Password: crose

Initial password, please change it immediately!

📅 Key Milestones

2026.05

• Integrated model training framework to support users in training local proprietary models.

2026.04

• High-quality prompts and dataset calls to large models for generating Node-RED flow services.

2026.03

• Platform basic functionality framework.