Skip to main content

Prerequisites

🧭 Overview​

Before deploying CSGHUB, ensure that your environment meets the following hardware and software requirements.

Properly configuring the runtime environment is the foundation for system stability and high performance.

⚙️ Hardware Requirements​

📦 Minimum Configuration (Test / Development Environment)​

Resource TypeSpecificationDescription
CPU4 CoreEnsures basic functionality
Memory8 GBSufficient for basic services
Storage200 GBIncludes system and data storage
Network1 GbpsSupports internal and external communication

🚀 Recommended Configuration (Production Environment)​

Resource TypeSpecificationDescription
CPU8 Core or moreSupports high concurrency and large-scale access
Memory16 GB or moreImproves performance and concurrency handling
Storage1 TB SSD or moreProvides high-speed data read/write
Network10 Gbps or moreSuitable for large file transfers and high traffic

🧠 Supported CPU Architectures​

  • ✅ AMD64 / x86-64: Fully supported
  • ✅ ARM64: Fully supported
  • ⚠️ Other architectures: Not supported

☸️ Kubernetes Requirements​

📦 Cluster Version​

  • Minimum Version: Kubernetes 1.28+
  • Recommended Version: Kubernetes 1.29+ or the latest stable version

🔧 Cluster Component Requirements

ComponentRequirementDescription
CNI PluginRequiredSupports Calico, Flannel, Cilium, etc.
CSI DriverRequiredFor dynamic volume management
Ingress ControllerRecommendedSupports Nginx or compatible implementations
LoadBalancerRecommended for productionCan use cloud LB or MetalLB

💾 Persistent Storage Requirements​

  • Must support Dynamic Provisioning
  • At least one StorageClass must exist
  • Storage plugin must comply with the CSI standard
  • Certain components (e.g., Dataflow, Csgship) require ReadWriteMany (RWX) capable storage, otherwise functionality may be limited

📘 Reference: Kubernetes Storage Configuration Guide

⛵ Helm Requirements​

🔢 Version Requirements​

  • Minimum Version: Helm 3.12+
  • Recommended Version: Helm 3.13+ or the latest stable version

🧠 Optional Component Requirements​

🤖 AI / ML Related Components​

Component NameRecommended VersionDescriptionAuto-Deployment Support
Knative Serving1.16.1+Model service autoscaling✅
Argo Workflowv3.5.12+Model evaluation & image building✅
LeaderWorkSetv0.6.1Multi-node, multi-GPU training✅
Nvidia Device PluginCUDA ≥12.1GPU acceleration support❌

📊 Monitoring & Logging Components​

Component NameRecommended VersionDescriptionAuto-Deployment Support
Prometheus2.47+Metrics collection✅
Grafana10.0+Data visualization✅
Loki2.9+Log collection & analysis✅

💽 Storage Components

Component NameRecommended VersionDescriptionAuto-Deployment Support
MinIORELEASE.2024-08-23T10-08-35ZObject storage✅
PostgreSQL15.0+Relational database✅
Redis7.2+Cache & session storage✅
MongoDB8.0+Logs & event storage✅

🌐 Network Requirements​

🔌 Required Open Ports​

PortProtocolDirectionPurpose
80TCPInboundHTTP access
443TCPInboundHTTPS access
22TCPInboundGit SSH access (optional)
30000–32767TCPInboundNodePort service range

🌍 Domain Requirements​

  • Must have a configurable valid domain name
  • Must support wildcard DNS (e.g., *.example.com)
  • DNS must resolve correctly to cluster entry nodes

🔒 Security Requirements​

🧾 Certificate Management​

EnvironmentRequirement
ProductionMust use a valid TLS certificate
TestCan use Let’s Encrypt (self-signed not recommended)

🧑‍💻 Access Control​

  • RBAC must be enabled
  • Ensure proper ServiceAccount permissions
  • Enabling NetworkPolicy for network isolation is recommended

✅ Environment Checklist​

Before deployment, run the following commands to verify that your cluster meets the requirements:

# Check Kubernetes version
kubectl version

# Check node status
kubectl get nodes

# Check StorageClass configuration
kubectl get storageclass

# Check Helm version
helm version

If any check fails, please resolve the issues before proceeding with the deployment.