timmy-home/uni-wizard/daemons/health_daemon.py

"""
Health Check Daemon for Uni-Wizard
Monitors VPS status and exposes health endpoint
"""

import json
import time
import threading
from http.server import HTTPServer, BaseHTTPRequestHandler
from datetime import datetime
from pathlib import Path
import sys

# Add parent to path
sys.path.insert(0, str(Path(__file__).parent.parent))

from harness import get_harness


class HealthCheckHandler(BaseHTTPRequestHandler):
    """HTTP handler for health endpoint"""
    
    def log_message(self, format, *args):
        # Suppress default logging
        pass
    
    def do_GET(self):
        """Handle GET requests"""
        if self.path == '/health':
            self.send_health_response()
        elif self.path == '/status':
            self.send_full_status()
        else:
            self.send_error(404)
    
    def send_health_response(self):
        """Send simple health check"""
        harness = get_harness()
        result = harness.execute("health_check")
        
        try:
            health_data = json.loads(result)
            status_code = 200 if health_data.get("overall") == "healthy" else 503
        except:
            status_code = 503
            health_data = {"error": "Health check failed"}
        
        self.send_response(status_code)
        self.send_header('Content-Type', 'application/json')
        self.end_headers()
        self.wfile.write(json.dumps(health_data).encode())
    
    def send_full_status(self):
        """Send full system status"""
        harness = get_harness()
        
        status = {
            "timestamp": datetime.now().isoformat(),
            "harness": json.loads(harness.get_status()),
            "system": json.loads(harness.execute("system_info")),
            "health": json.loads(harness.execute("health_check"))
        }
        
        self.send_response(200)
        self.send_header('Content-Type', 'application/json')
        self.end_headers()
        self.wfile.write(json.dumps(status, indent=2).encode())


class HealthDaemon:
    """
    Health monitoring daemon.
    
    Runs continuously, monitoring:
    - System resources
    - Service status
    - Inference endpoint
    
    Exposes:
    - HTTP endpoint on port 8082
    - JSON status file at ~/timmy/logs/health.json
    """
    
    def __init__(self, port: int = 8082, check_interval: int = 60):
        self.port = port
        self.check_interval = check_interval
        self.running = False
        self.server = None
        self.monitor_thread = None
        self.last_health = None
        
        # Ensure log directory exists
        self.log_path = Path.home() / "timmy" / "logs"
        self.log_path.mkdir(parents=True, exist_ok=True)
        self.health_file = self.log_path / "health.json"
    
    def start(self):
        """Start the health daemon"""
        self.running = True
        
        # Start HTTP server
        self.server = HTTPServer(('127.0.0.1', self.port), HealthCheckHandler)
        server_thread = threading.Thread(target=self.server.serve_forever)
        server_thread.daemon = True
        server_thread.start()
        
        # Start monitoring loop
        self.monitor_thread = threading.Thread(target=self._monitor_loop)
        self.monitor_thread.daemon = True
        self.monitor_thread.start()
        
        print(f"Health daemon started on http://127.0.0.1:{self.port}")
        print(f"  - /health  - Quick health check")
        print(f"  - /status  - Full system status")
        print(f"Health file: {self.health_file}")
    
    def stop(self):
        """Stop the health daemon"""
        self.running = False
        if self.server:
            self.server.shutdown()
        print("Health daemon stopped")
    
    def _monitor_loop(self):
        """Background monitoring loop"""
        while self.running:
            try:
                self._update_health_file()
                time.sleep(self.check_interval)
            except Exception as e:
                print(f"Monitor error: {e}")
                time.sleep(5)
    
    def _update_health_file(self):
        """Update the health status file"""
        harness = get_harness()
        
        try:
            health_result = harness.execute("health_check")
            system_result = harness.execute("system_info")
            
            status = {
                "timestamp": datetime.now().isoformat(),
                "health": json.loads(health_result),
                "system": json.loads(system_result)
            }
            
            self.health_file.write_text(json.dumps(status, indent=2))
            self.last_health = status
            
        except Exception as e:
            print(f"Failed to update health file: {e}")


def main():
    """Run the health daemon"""
    import signal
    
    daemon = HealthDaemon()
    
    def signal_handler(sig, frame):
        print("\nShutting down...")
        daemon.stop()
        sys.exit(0)
    
    signal.signal(signal.SIGINT, signal_handler)
    signal.signal(signal.SIGTERM, signal_handler)
    
    daemon.start()
    
    # Keep main thread alive
    try:
        while True:
            time.sleep(1)
    except KeyboardInterrupt:
        daemon.stop()


if __name__ == "__main__":
    main()