Installation Overview
📘 Overview
CSGHUB is an integrated AI R&D and collaboration platform that supports the full lifecycle of model management → inference → fine-tuning → application deployment.
To meet various deployment scenarios and scales, CSGHUB provides multiple installation methods:
| Deployment Method | Use Case | Complexity | Highlights |
|---|---|---|---|
| Helm Chart (Recommended) | Production / Cloud-native | ⭐⭐⭐⭐ | High availability, multi-node support, auto-scaling |
| Docker Compose | Testing / Development / Single node | ⭐⭐⭐ | Lightweight and easy to start |
🧱 Component Architecture
CSGHUB consists of multiple core and optional components:
| Module Name | Type | Description |
|---|---|---|
| accounting | Core Service | User billing, compute quota, and usage tracking |
| casdoor | Auth Service | Authentication and identity center (OAuth / OIDC) |
| dataviewer | Application | Dataset browsing and visualization |
| gitaly | Storage Service | Git repository backend for code and model versions |
| gitlab_shell | Helper Service | Git SSH access layer for code push/pull operations |
| loki | Logging Service | Centralized log collection and querying |
| minio | Storage Service | Object storage for models, datasets, and artifacts |
| mirror_lfs | Sync Service | Manages LFS (large file) mirror synchronization |
| mirror_repo | Sync Service | Handles repository mirroring and remote sync |
| nats | Message Middleware | Event-driven messaging and async task queue |
| nginx | Gateway | Web traffic entry and reverse proxy |
| notification | Message Service | System notifications and async message dispatching |
| portal | Frontend | Web UI for user interaction |
| postgresql | Database | Core data store for models, users, and tasks |
| prometheus | Monitoring | System and component metric collection |
| redis | Cache Service | Session caching and queue backend |
| registry | Image Registry | Stores container and model images (OCI compatible) |
| rproxy | Internal Proxy | Internal routing and unified request proxy |
| server | Core Backend | Main service for scheduling, API, and orchestration |
| temporal | Workflow Service | Distributed task orchestration and workflow engine |
| temporal_ui | Management UI | Temporal workflow web console |
| temporal_worker | Worker Node | Task execution and async process runner |
| user | User Service | User, organization, and permissions management |
| csgship | AI Assistant API | Backend service for the CodeSouler IDE integration |
🧩 System Layers
| Layer | Examples | Purpose |
|---|---|---|
| Gateway Layer | nginx, rproxy | External access and traffic routing |
| Core Service Layer | server, portal, user, accounting, notification, dataviewer | Main platform functionality |
| Storage & Infrastructure | postgresql, redis, minio, registry, gitaly, gitlab-shell, casdoor | Data persistence and management |
| Workflow & Queue Layer | temporal, temporal_worker, nats, mirror_* | Workflow orchestration and messaging |
| Auxiliary Layer | loki, prometheus, temporal_ui | Logging, monitoring, visualization |
| AI Assistant Layer | csgship | IDE integration backend |
⚙️ System Requirements
✅ Hardware Requirements (Minimum / Recommended)
| Resource | Minimum | Recommended |
|---|---|---|
| CPU / Memory | 4 cores / 8GB | 8 cores / 16GB |
| Storage | 50 GB | 200 GB+ |
| Architecture | AMD64 / ARM64 | Supports x86 and ARM |
✅ Software Requirements
| Component | Version |
|---|---|
| Docker Engine | ≥ 20.10 |
| Docker Compose | ≥ 2.0 |
| Kubernetes (optional) | ≥ 1.28 |
| Helm | ≥ 3.12 |
🧩 Deployment Options
🪄 1️⃣ Quick Start (Recommended for New Users)
Ideal for demo environments or feature validation
-
Method: Docker Compose
-
Startup time: ~2 minutes
-
Features: Single-node, one-click startup
-
Limitations:
- Does not include the Runner service (requires Kubernetes)
- Does not include the Dataflow service (available via Docker Compose / Helm)
☁️ 2️⃣ Standard Deployment (Production Recommended)
Suitable for production and enterprise clusters
-
Method: Helm Chart
-
Highlights:
- Multi-node and high-availability support
- Integrates with external DBs, Redis, and object storage
- Compatible with major cloud providers (Alibaba Cloud, Tencent Cloud, AWS, etc.)
🌍 External Resource Configuration
For production environments, it is strongly recommended to use externally managed services (databases, caches, storage) to improve reliability and scalability.
| Resource Type | Recommended Solution |
|---|---|
| PostgreSQL | Alibaba Cloud RDS / AWS RDS / Cloud SQL |
| Redis | Alibaba Cloud Redis / AWS ElastiCache |
| Object Storage | Alibaba Cloud OSS / AWS S3 / External MinIO |
| Registry | Private registry (Harbor / GitLab Registry) |
💬 Support & Feedback
- Official Website: https://opencsg.com
- GitHub Issues: https://github.com/OpenCSGs/csghub-charts/issues