DocsServer Node

CyxWiz Server Node

The Server Node (also called "miner") is the distributed compute worker that executes ML training jobs. It can run in GUI mode for interactive use or as a background daemon.

Overview

The Server Node provides:

Job Execution

Train models using GPU/CPU

Hardware Monitoring

Track resource utilization

OpenAI-Compatible API

Serve models via REST

Central Server Communication

GUI and Daemon Modes

Flexible deployment

Running Modes

GUI Mode (Default)

Interactive application with visual dashboard:

# Windows
.\cyxwiz-server-node.exe

# Linux
./cyxwiz-server-node

Daemon Mode

Background service for production:

# Windows
.\cyxwiz-server-node.exe --daemon

# Linux
./cyxwiz-server-node --daemon

TUI Mode

Terminal-based interface:

./cyxwiz-server-node --tui

GUI Interface

+------------------------------------------------------------------+
|  CyxWiz Server Node v0.1.0                      [_] [x]          |
+------------------------------------------------------------------+
|  File  View  Settings  Help                                       |
+------------------------------------------------------------------+
|                                                                   |
|  DASHBOARD                                                        |
|  +----------------------------+  +----------------------------+   |
|  | STATUS: Online             |  | JOBS                       |   |
|  | Node ID: abc123...         |  | Active: 2                  |   |
|  | Uptime: 12h 34m            |  | Queued: 5                  |   |
|  | Central: Connected         |  | Completed: 156             |   |
|  +----------------------------+  +----------------------------+   |
|                                                                   |
|  HARDWARE                                                         |
|  +-----------------------------------------------------------+   |
|  | CPU:  [============        ] 58%   Cores: 8               |   |
|  | RAM:  [========            ] 42%   12.4/32 GB             |   |
|  | GPU:  [================    ] 78%   RTX 4060 8GB           |   |
|  | VRAM: [==============      ] 71%   5.7/8 GB               |   |
|  +-----------------------------------------------------------+   |
|                                                                   |
+------------------------------------------------------------------+

GUI Panels

Panel	Description
Dashboard	Overview of node status
Hardware	CPU, GPU, memory monitoring
Jobs	Active and queued jobs
Deployment	Model deployment management
API Keys	Manage API access keys
Wallet	CYXWIZ token management

OpenAI-Compatible API

The Server Node can serve deployed models via a REST API compatible with OpenAI's API format.

Endpoints

Method	Path	Description
GET	/v1/models	List available models
POST	/v1/chat/completions	Chat completion
POST	/v1/completions	Text completion
POST	/v1/embeddings	Generate embeddings

Example Usage

# List models
curl http://localhost:8000/v1/models

# Chat completion
curl http://localhost:8000/v1/chat/completions \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deployed-model-1",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Configuration

Located at ~/.cyxwiz/server-node/config.toml:

[node]
name = "my-node"
max_concurrent_jobs = 3

[central_server]
address = "localhost:50051"
heartbeat_interval_seconds = 10
reconnect_delay_seconds = 5

[devices]
# Specific GPU selection (empty = all)
gpu_ids = []
# Maximum memory per job
max_gpu_memory_mb = 8000

[api]
enabled = true
port = 8000

[logging]
level = "info"
file = "server-node.log"

Requirements

GPU: NVIDIA with CUDA 11+ (optional)
RAM: 16GB minimum

Storage: 100GB for models/datasets
Network: Stable connection to Central Server