EVOSEAL Phase 3 Deployment Guide¶

Overview¶

This guide covers the complete deployment of EVOSEAL Phase 3 Bidirectional Continuous Evolution System, from initial setup to production operation with systemd integration.

Prerequisites¶

System Requirements¶

OS: Linux (Ubuntu 20.04+ recommended)
Python: 3.10 or higher
Memory: 8GB RAM minimum (32GB recommended for GPU fine-tuning)
Storage: 50GB free space (for models and evolution data)
Network: Internet access for initial setup

Hardware Requirements¶

CPU: Multi-core processor (6+ cores recommended)
GPU: Optional but recommended (RTX 4090 or equivalent for full fine-tuning)
Disk: SSD recommended for better performance

Installation¶

Step 1: Clone Repository¶

git clone https://github.com/SHA888/EVOSEAL
cd EVOSEAL

Step 2: Python Environment Setup¶

# Create virtual environment
python3 -m venv .venv
source .venv/bin/activate

# Install core dependencies
pip install -e .

# Install Phase 3 specific dependencies
pip install aiohttp aiohttp-cors pydantic-settings

# Install in development mode
pip install -e .

Step 3: Ollama and Local Model Setup¶

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull the coder and reviewer models (CPU-friendly prompt-level path).
# EVOSEAL auto-discovers these from Ollama by family, so any coder/reviewer
# model works; these are the defaults:
ollama pull deepseek-coder-v2:16b-lite-instruct-q8_0   # coder (~16GB)
ollama pull qwen2.5-coder:7b-instruct-q6_K             # reviewer (~6GB)

# Start Ollama service
ollama serve &

# Verify installation
ollama list
curl http://localhost:11434/api/tags

Step 4: Configuration¶

# Copy environment template (if exists)
cp .env.example .env

# Edit configuration as needed
nano .env

Step 5: Health Check¶

# Run comprehensive health check
python3 scripts/run_phase3_continuous_evolution.py --health-check

Deployment Options¶

Option 1: Development/Testing Deployment¶

For development and testing purposes:

# Start Phase 3 system directly
python3 scripts/run_phase3_continuous_evolution.py --verbose

# Access dashboard
open http://localhost:9613

Option 2: Production Deployment with systemd¶

For production environments (recommended):

Enable User Linger¶

# Enable user linger for auto-start on boot
sudo loginctl enable-linger $USER

Service Configuration¶

The systemd service is pre-configured at:

~/.config/systemd/user/evoseal.service

Start Production Service¶

# Reload systemd configuration
systemctl --user daemon-reload

# Enable and start service
systemctl --user enable evoseal.service
systemctl --user start evoseal.service

# Verify service status
systemctl --user status evoseal.service

# Access production dashboard
open http://localhost:9613

Configuration¶

Environment Variables¶

Key environment variables for Phase 3:

# EVOSEAL Core
export EVOSEAL_ROOT="/home/kade/EVOSEAL"
export EVOSEAL_LOGS="/home/kade/EVOSEAL/logs"
export PYTHONPATH="/home/kade/EVOSEAL:/home/kade/EVOSEAL/SEAL"

# Phase 3 Specific
export EVOSEAL_DASHBOARD_PORT="9613"
export EVOSEAL_EVOLUTION_INTERVAL="3600"  # 1 hour
export EVOSEAL_TRAINING_INTERVAL="1800"   # 30 minutes
export EVOSEAL_MIN_SAMPLES="50"

# Ollama Configuration
# Models are auto-discovered from Ollama per role; override only to pin a tag.
export OLLAMA_HOST="http://localhost:11434"
export EVOSEAL_CODER_MODEL="deepseek-coder-v2:16b-lite-instruct-q8_0"
export EVOSEAL_REVIEWER_MODEL="qwen2.5-coder:7b-instruct-q6_K"

Command Line Options¶

Phase 3 system supports extensive configuration:

python3 scripts/run_phase3_continuous_evolution.py \
  --port=9613 \                        # Dashboard port
  --evolution-interval=3600 \          # Evolution check interval (seconds)
  --training-interval=1800 \           # Training check interval (seconds)
  --min-samples=50 \                   # Minimum samples for training
  --verbose                            # Enable verbose logging

systemd Service Configuration¶

Edit service configuration if needed:

# Edit service file
nano ~/.config/systemd/user/evoseal.service

# Key configuration sections:
# - ExecStart: Main command and arguments
# - Environment: Environment variables
# - Restart: Restart behavior
# - Logging: Log file locations

# Reload after changes
systemctl --user daemon-reload
systemctl --user restart evoseal.service

Monitoring and Management¶

Real-time Dashboard¶

Access the monitoring dashboard: - Development: http://localhost:8081 - Production: http://localhost:8081

Dashboard features: - Service status and uptime - Evolution metrics and progress - Training status and model versions - Performance analytics - Live activity log with WebSocket updates

Service Management¶

# Check service status
systemctl --user status evoseal.service

# View real-time logs
journalctl --user -fu evoseal.service

# Restart service
systemctl --user restart evoseal.service

# Stop/start service
systemctl --user stop evoseal.service
systemctl --user start evoseal.service

Log Management¶

Logs are written to multiple locations:

# Application logs
tail -f /home/kade/EVOSEAL/logs/phase3_continuous_evolution.log

# Service logs (stdout)
tail -f /home/kade/EVOSEAL/logs/evoseal.log

# Error logs (stderr)
tail -f /home/kade/EVOSEAL/logs/evoseal-error.log

# systemd journal
journalctl --user -fu evoseal.service

Data Management¶

Data Directories¶

Phase 3 creates several data directories:

data/continuous_evolution/
├── evolution_data/          # Phase 1: Evolution data collection
├── bidirectional/           # Phase 3: Bidirectional evolution data
│   ├── training/           # Training data and models
│   │   └── versions/       # Model versions
│   └── evolution_report_*.json  # Evolution reports
└── logs/                   # Application logs

Backup Strategy¶

# Backup evolution data
tar -czf evoseal_data_backup_$(date +%Y%m%d).tar.gz data/

# Backup configuration
cp ~/.config/systemd/user/evoseal.service ~/evoseal_service_backup.service
cp .env ~/evoseal_env_backup.env

# Backup logs (optional)
tar -czf evoseal_logs_backup_$(date +%Y%m%d).tar.gz logs/

Data Cleanup¶

# Clean old evolution reports (older than 30 days)
find data/continuous_evolution/bidirectional/ -name "evolution_report_*.json" -mtime +30 -delete

# Clean old model versions (keep last 10)
cd data/continuous_evolution/bidirectional/training/versions/
ls -t | tail -n +11 | xargs rm -rf

# Rotate logs
find logs/ -name "*.log" -mtime +7 -exec gzip {} \;

Security¶

Service Security¶

User Service: Runs as user service, no root privileges required
NoNewPrivileges: Prevents privilege escalation
Local Access: Dashboard only accessible on localhost
Environment Isolation: Controlled environment variables

Network Security¶

Local Operation: All processing local, no external API calls
Ollama Local: Model inference entirely local
Dashboard Binding: Dashboard only binds to localhost interface
No External Exposure: No services exposed to external networks

Data Security¶

Local Storage: All data stored locally
File Permissions: Proper file permissions on data directories
Log Rotation: Automatic log rotation to prevent disk space issues

Performance Optimization¶

System Optimization¶

# Increase file descriptor limits
echo "* soft nofile 65536" | sudo tee -a /etc/security/limits.conf
echo "* hard nofile 65536" | sudo tee -a /etc/security/limits.conf

# Optimize Python performance
export PYTHONOPTIMIZE=1
export PYTHONDONTWRITEBYTECODE=1

Resource Monitoring¶

# Monitor system resources
htop

# Monitor disk usage
df -h
du -sh data/

# Monitor memory usage
free -h

# Monitor service resource usage
systemctl --user show evoseal.service --property=MemoryCurrent,CPUUsageNSec

Performance Tuning¶

# Adjust evolution intervals for performance
python3 scripts/run_phase3_continuous_evolution.py \
  --evolution-interval=7200 \    # Check every 2 hours
  --training-interval=3600 \     # Check training every hour
  --min-samples=100              # Require more samples before training

Troubleshooting¶

Common Issues¶

Port Already in Use¶

# Check what's using the port
netstat -tlnp | grep :8081

# Kill process using port
sudo fuser -k 8081/tcp

# Or use different port
python3 scripts/run_phase3_continuous_evolution.py --port=8082

Ollama Not Running¶

# Check Ollama status
curl http://localhost:11434/api/tags

# Start Ollama
ollama serve &

# Check that a coder-family model is available
ollama list | grep -iE 'coder|code'

Service Won't Start¶

# Check service status
systemctl --user status evoseal.service

# View detailed logs
journalctl --user -xeu evoseal.service

# Run health check
python3 scripts/run_phase3_continuous_evolution.py --health-check

Permission Issues¶

# Check file permissions
ls -la ~/.config/systemd/user/evoseal.service
ls -la /home/kade/EVOSEAL/logs/

# Fix permissions if needed
chmod 644 ~/.config/systemd/user/evoseal.service
chmod -R 755 /home/kade/EVOSEAL/logs/

Diagnostic Commands¶

# Complete system diagnostic
echo "=== EVOSEAL Phase 3 Diagnostic ==="
echo "Service Status:"
systemctl --user status evoseal.service

echo -e "\nOllama Status:"
curl -s http://localhost:11434/api/tags | jq .

echo -e "\nDisk Usage:"
df -h /home/kade/EVOSEAL

echo -e "\nMemory Usage:"
free -h

echo -e "\nRecent Logs:"
journalctl --user -u evoseal.service --lines=10 --no-pager

Maintenance¶

Regular Maintenance Tasks¶

Daily¶

Check service status
Monitor dashboard for issues
Review error logs

Weekly¶

Check disk usage
Review evolution progress
Update system packages

Monthly¶

Backup evolution data
Clean old logs and reports
Update EVOSEAL codebase
Review performance metrics

Update Procedure¶

# Stop service
systemctl --user stop evoseal.service

# Backup current state
tar -czf evoseal_backup_$(date +%Y%m%d).tar.gz data/ logs/ .env

# Update codebase
git pull origin main

# Update dependencies
pip install -e .

# Run health check
python3 scripts/run_phase3_continuous_evolution.py --health-check

# Restart service
systemctl --user start evoseal.service

# Verify operation
systemctl --user status evoseal.service

Scaling and High Availability¶

Horizontal Scaling¶

For larger deployments:

# Run multiple instances on different ports
python3 scripts/run_phase3_continuous_evolution.py --port=8081 &
python3 scripts/run_phase3_continuous_evolution.py --port=8082 &

# Use load balancer (nginx example)
upstream evoseal_backend {
    server localhost:8081;
    server localhost:8082;
}

High Availability Setup¶

# Set up multiple Ollama instances
ollama serve --port 11434 &
ollama serve --port 11435 &

# Configure provider fallback in EVOSEAL
# Edit configuration to include multiple Ollama endpoints

Conclusion¶

This deployment guide provides comprehensive instructions for setting up EVOSEAL Phase 3 in both development and production environments. The systemd integration ensures reliable operation with automatic startup, restart capabilities, and comprehensive monitoring.

For additional support and troubleshooting, refer to: - Phase 3 Documentation - systemd Integration Guide - Main README

The Phase 3 system is designed for continuous operation and will automatically evolve and improve both EVOSEAL and its local coding model through the bidirectional evolution loop.