The gateway command manages the lifecycle of all NeuraTrade services (Backend API, CCXT Service, Telegram Service).
Start Services
Start all NeuraTrade services:
./bin/neuratrade gateway start
Example Output
🚀 Starting NeuraTrade Gateway...
📁 NeuraTrade Home: /home/user/.neuratrade
⚙️ Config File: /home/user/.neuratrade/config.json
🌐 Backend Port: 8080 (public)
🔌 CCXT Port: 3001 (internal, bound to 127.0.0.1)
📞 Telegram Port: 3002 (internal, bound to 127.0.0.1)
🛡️ Supervised Mode: false
⏱️ Health Timeout: 90s
📊 Starting CCXT Service...
✅ CCXT Service started
🔧 Starting Backend API...
✅ Backend API started
📞 Starting Telegram Service...
✅ Telegram Service started
🎉 All services started successfully!
📡 Backend API: http://localhost:8080
🏥 Health Check: http://localhost:8080/health
Press Ctrl+C to stop all services
Flags
--supervised
Enable supervised mode to keep the gateway running even if health checks fail during startup:
./bin/neuratrade gateway start --supervised
Supervised Mode allows services to warm up gradually. The gateway will start services and continue running even if initial health probes fail, giving backend and Telegram services time to initialize.Set via flag or environment variable:export NEURATRADE_GATEWAY_SUPERVISED=true
Environment Variables
| Variable | Description | Default |
|---|
SERVER_PORT | Backend API port | 8080 |
CCXT_PORT | CCXT service port | 3001 |
TELEGRAM_PORT | Telegram service port | 3002 |
BIND_HOST | Service bind address | 127.0.0.1 |
NEURATRADE_GATEWAY_SUPERVISED | Enable supervised mode | false |
NEURATRADE_GATEWAY_HEALTH_TIMEOUT_SECONDS | Health check timeout | 90 |
ADMIN_API_KEY | Admin API key (auto-generated if missing) | - |
JWT_SECRET | JWT signing secret (auto-generated if missing) | - |
SQLITE_PATH | SQLite database path | ~/.neuratrade/data/neuratrade.db |
TELEGRAM_BOT_TOKEN | Telegram bot token | From config |
AI_API_KEY | AI provider API key | From config |
Service Startup Order
- CCXT Service - Exchange data bridge
- Backend API - Core API server (waits for health check)
- Telegram Service - Bot interface (waits for health check)
Health Monitoring
The gateway continuously monitors service health every 15 seconds and updates the runtime state:
- healthy - All services running and responding to health checks
- warming - Services started but health checks not yet passing
- degraded - Some services unhealthy
- down - All services stopped
Security Note: The backend binds to 0.0.0.0 (all interfaces) to accept external connections, while CCXT and Telegram services bind to 127.0.0.1 (localhost only) for internal communication.Ensure firewall rules protect the backend port in production.
Stop Services
Stop all running NeuraTrade services:
./bin/neuratrade gateway stop
Example Output
🛑 Stopping NeuraTrade services...
✅ Backend API: Stopped (PID: 12345)
✅ CCXT Service: Stopped (PID: 12346)
✅ Telegram Service: Stopped (PID: 12347)
✅ Stopped 3 service(s)
How It Works
- Reads PID files from
~/.neuratrade/pids/
- Validates process is still running and matches expected pattern
- Sends
SIGTERM signal for graceful shutdown
- Removes PID files
Force Stop
If gateway stop doesn’t work, manually kill processes:
pkill -f neuratrade-server
pkill -f ccxt-service
pkill -f telegram-service
Status Check
Check the status of all services:
./bin/neuratrade gateway status
Example Output
📊 NeuraTrade Service Status
============================
Runtime Mode: HEALTHY
Last Update: 2026-03-03T10:30:00Z
✅ Backend API: Running (PID: 12345)
✅ CCXT Service: Running (PID: 12346)
✅ Telegram Service: Running (PID: 12347)
🚪 Health Check: http://127.0.0.1:8080/health
✅ Backend Status: healthy
Service Health:
✓ database: healthy
✓ redis: healthy
✓ ccxt: healthy
✓ telegram: healthy
Status When Services Are Down
📊 NeuraTrade Service Status
============================
❌ Backend API: Not running
❌ CCXT Service: Not running
❌ Telegram Service: Not running
🚪 Health Check: http://127.0.0.1:8080/health
⚠️ Backend health probe failed: connection refused
Gateway runtime mode is DOWN (services may still be warming).
Make sure the backend is running:
neuratrade gateway start
Process Management
PID Files
PID files track running processes:
~/.neuratrade/pids/backend.pid~/.neuratrade/pids/ccxt.pid~/.neuratrade/pids/telegram.pid~/.neuratrade/pids/gateway-state.json
Stale PID Files
The gateway stop command automatically detects and removes stale PID files when:
- Process no longer exists
- Process doesn’t match expected pattern (e.g.,
neuratrade-server)
Manual Cleanup
Remove PID files if processes were terminated externally:
rm ~/.neuratrade/pids/*.pid
rm ~/.neuratrade/pids/gateway-state.json
Service Logs
View service logs in real-time:
# Backend API
tail -f ~/.neuratrade/logs/backend.log
# CCXT Service
tail -f ~/.neuratrade/logs/ccxt.log
# Telegram Service
tail -f ~/.neuratrade/logs/telegram.log
Graceful Shutdown
When running gateway start interactively, press Ctrl+C for graceful shutdown:
^C
🛑 Shutting down services...
✅ All services stopped
SIGTERM to all services, allowing them to:
- Close database connections
- Flush logs
- Complete in-flight requests
- Clean up resources
Troubleshooting
Services Won’t Start
Check for port conflicts:
lsof -i :8080 # Backend
lsof -i :3001 # CCXT
lsof -i :3002 # Telegram
Health Checks Failing
Increase the health timeout:
export NEURATRADE_GATEWAY_HEALTH_TIMEOUT_SECONDS=120
./bin/neuratrade gateway start
./bin/neuratrade gateway start --supervised
Services Start Then Crash
Check service logs:
tail -100 ~/.neuratrade/logs/backend.log
tail -100 ~/.neuratrade/logs/ccxt.log
tail -100 ~/.neuratrade/logs/telegram.log
Can’t Stop Services
Force kill all processes:
pkill -9 -f neuratrade-server
pkill -9 -f ccxt-service
pkill -9 -f telegram-service
rm ~/.neuratrade/pids/*.pid