claude-mpm 5.0.2__py3-none-any.whl → 5.4.3__py3-none-any.whl

claude_mpm/services/monitor/daemon.py

@@ -44,6 +44,7 @@ class UnifiedMonitorDaemon:
         daemon_mode: bool = False,
         pid_file: Optional[str] = None,
         log_file: Optional[str] = None,
+        enable_hot_reload: bool = False,
     ):
         """Initialize the unified monitor daemon.
 
@@ -53,10 +54,12 @@ class UnifiedMonitorDaemon:
             daemon_mode: Whether to run as background daemon
             pid_file: Path to PID file for daemon mode
             log_file: Path to log file for daemon mode
+            enable_hot_reload: Enable file watching and hot reload for development
         """
         self.host = host
         self.port = port
         self.daemon_mode = daemon_mode
+        self.enable_hot_reload = enable_hot_reload
         self.logger = get_logger(__name__)
 
         # Use new consolidated DaemonManager for all daemon operations
@@ -75,7 +78,9 @@ class UnifiedMonitorDaemon:
         )
 
         # Core server
-        self.server = UnifiedMonitorServer(host=host, port=port)
+        self.server = UnifiedMonitorServer(
+            host=host, port=port, enable_hot_reload=enable_hot_reload
+        )
 
         # Health monitoring
         self.health_monitor = HealthMonitor(port=port)
@@ -88,11 +93,12 @@ class UnifiedMonitorDaemon:
         self.shutdown_event = threading.Event()
 
     def _get_default_pid_file(self) -> str:
-        """Get default PID file path."""
+        """Get default PID file path with port number to support multiple daemons."""
         project_root = Path.cwd()
         claude_mpm_dir = project_root / ".claude-mpm"
         claude_mpm_dir.mkdir(exist_ok=True)
-        return str(claude_mpm_dir / "monitor-daemon.pid")
+        # Include port in filename to support multiple daemon instances
+        return str(claude_mpm_dir / f"monitor-daemon-{self.port}.pid")
 
     def start(self, force_restart: bool = False) -> bool:
         """Start the unified monitor daemon.
@@ -145,16 +151,26 @@ class UnifiedMonitorDaemon:
         if self.daemon_manager.is_running():
             existing_pid = self.daemon_manager.get_pid()
             if not force_restart:
-                self.logger.warning(f"Daemon already running with PID {existing_pid}")
+                msg = f"Daemon already running on port {self.port} with PID {existing_pid}"
+                self.logger.warning(msg)
+                # If we're in subprocess mode, this is an error - we should have cleaned up
+                if os.environ.get("CLAUDE_MPM_SUBPROCESS_DAEMON") == "1":
+                    self.logger.error(
+                        f"SUBPROCESS ERROR: {msg} - This should not happen in subprocess mode!"
+                    )
                 return False
             # Force restart was already handled above
 
         # Check for our service on the port
         is_ours, pid = self.daemon_manager.is_our_service()
         if is_ours and pid and not force_restart:
-            self.logger.warning(
-                f"Our service already running on port {self.port} (PID: {pid})"
-            )
+            msg = f"Our service already running on port {self.port} (PID: {pid})"
+            self.logger.warning(msg)
+            # If we're in subprocess mode, this is an error - we should have cleaned up
+            if os.environ.get("CLAUDE_MPM_SUBPROCESS_DAEMON") == "1":
+                self.logger.error(
+                    f"SUBPROCESS ERROR: {msg} - This should not happen in subprocess mode!"
+                )
             return False
 
         # Use subprocess approach for clean daemon startup (v4.2.40)
@@ -169,20 +185,29 @@ class UnifiedMonitorDaemon:
         if os.environ.get("CLAUDE_MPM_SUBPROCESS_DAEMON") == "1":
             # We're in a subprocess started by start_daemon_subprocess
             # We need to write the PID file ourselves since parent didn't
-            self.logger.info("Running in subprocess daemon mode, writing PID file")
+            self.logger.info(f"Running in subprocess daemon mode on port {self.port}")
+            self.logger.info(f"Subprocess PID: {os.getpid()}")
+            self.logger.info(f"PID file path: {self.daemon_manager.pid_file}")
             self.daemon_manager.write_pid_file()
+            self.logger.info("PID file written successfully")
 
             # Setup signal handlers for graceful shutdown
             self._setup_signal_handlers()
+            self.logger.info("Signal handlers configured")
 
             # Start the server (this will run until shutdown)
+            self.logger.info("Starting server in subprocess mode...")
             try:
                 result = self._run_server()
                 if not result:
                     self.logger.error("Failed to start server in subprocess mode")
+                    return result
+                self.logger.info("Server started successfully in subprocess mode")
                 return result
             except Exception as e:
-                self.logger.error(f"Server startup exception in subprocess: {e}")
+                self.logger.error(
+                    f"Server startup exception in subprocess: {e}", exc_info=True
+                )
                 raise
         else:
             # Legacy fork approach (kept for compatibility but not used by default)
@@ -490,7 +515,9 @@ class UnifiedMonitorDaemon:
 
         # Recreate the server and health monitor after stop() sets them to None
         self.logger.info(f"Recreating server components for {self.host}:{self.port}")
-        self.server = UnifiedMonitorServer(host=self.host, port=self.port)
+        self.server = UnifiedMonitorServer(
+            host=self.host, port=self.port, enable_hot_reload=self.enable_hot_reload
+        )
         self.health_monitor = HealthMonitor(port=self.port)
 
         # Reset the shutdown event for the new run

claude_mpm/services/monitor/daemon_manager.py

@@ -34,6 +34,11 @@ from typing import Optional, Tuple
 from ...core.enums import OperationResult
 from ...core.logging_config import get_logger
 
+# Exit code constants for signal handling
+EXIT_NORMAL = 0
+EXIT_SIGKILL = 137  # 128 + SIGKILL(9) - forced termination
+EXIT_SIGTERM = 143  # 128 + SIGTERM(15) - graceful shutdown
+
 
 class DaemonManager:
     """Centralized manager for all daemon lifecycle operations.
@@ -80,18 +85,20 @@ class DaemonManager:
         self.startup_status_file = None
 
     def _get_default_pid_file(self) -> Path:
-        """Get default PID file path."""
+        """Get default PID file path with port number to support multiple daemons."""
         project_root = Path.cwd()
         claude_mpm_dir = project_root / ".claude-mpm"
         claude_mpm_dir.mkdir(exist_ok=True)
-        return claude_mpm_dir / "monitor-daemon.pid"
+        # Include port in filename to support multiple daemon instances
+        return claude_mpm_dir / f"monitor-daemon-{self.port}.pid"
 
     def _get_default_log_file(self) -> Path:
-        """Get default log file path."""
+        """Get default log file path with port number to support multiple daemons."""
         project_root = Path.cwd()
         claude_mpm_dir = project_root / ".claude-mpm"
         claude_mpm_dir.mkdir(exist_ok=True)
-        return claude_mpm_dir / "monitor-daemon.log"
+        # Include port in filename to support multiple daemon instances
+        return claude_mpm_dir / f"monitor-daemon-{self.port}.log"
 
     def cleanup_port_conflicts(self, max_retries: int = 3) -> bool:
         """Clean up any processes using the daemon port.
@@ -471,6 +478,55 @@ class DaemonManager:
 
         return None
 
+    def _verify_daemon_health(self, max_attempts: int = 3) -> bool:
+        """Verify daemon is healthy by checking HTTP health endpoint.
+
+        Args:
+            max_attempts: Maximum number of connection attempts
+
+        Returns:
+            True if health check passes, False otherwise
+        """
+        try:
+            import requests
+
+            for attempt in range(max_attempts):
+                try:
+                    # Try to connect to health endpoint
+                    response = requests.get(
+                        f"http://{self.host}:{self.port}/health", timeout=2
+                    )
+
+                    if response.status_code == 200:
+                        self.logger.debug(
+                            f"Health check passed on attempt {attempt + 1}/{max_attempts}"
+                        )
+                        return True
+
+                    self.logger.debug(
+                        f"Health check returned status {response.status_code} on attempt {attempt + 1}/{max_attempts}"
+                    )
+
+                except requests.exceptions.RequestException as e:
+                    self.logger.debug(
+                        f"Health check attempt {attempt + 1}/{max_attempts} failed: {e}"
+                    )
+
+                # Wait before retry (except on last attempt)
+                if attempt < max_attempts - 1:
+                    time.sleep(1)
+
+            self.logger.debug(f"Health check failed after {max_attempts} attempts")
+            return False
+
+        except ImportError:
+            # requests not available, skip health check
+            self.logger.debug("requests library not available, skipping health check")
+            return True
+        except Exception as e:
+            self.logger.debug(f"Health check error: {e}")
+            return False
+
     def start_daemon(self, force_restart: bool = False) -> bool:
         """Start the daemon with automatic cleanup and retry.
 
@@ -505,7 +561,7 @@ class DaemonManager:
             # Use subprocess for clean daemon startup (v4.2.40)
             # This avoids fork() issues with Python threading
             if self.use_subprocess_daemon():
-                return self.start_daemon_subprocess()
+                return self.start_daemon_subprocess(force_restart=force_restart)
             # Fallback to traditional fork (kept for compatibility)
             return self.daemonize()
 
@@ -523,12 +579,15 @@ class DaemonManager:
         # Otherwise, use subprocess for monitor daemon to avoid threading issues
         return True
 
-    def start_daemon_subprocess(self) -> bool:
+    def start_daemon_subprocess(self, force_restart: bool = False) -> bool:
         """Start daemon using subprocess.Popen for clean process isolation.
 
         This avoids all the fork() + threading issues by starting the monitor
         in a completely fresh process with no inherited threads or locks.
 
+        Args:
+            force_restart: Whether this is a force restart (helps interpret exit codes)
+
         Returns:
             True if daemon started successfully
         """
@@ -588,36 +647,90 @@ class DaemonManager:
                 pid = process.pid
                 self.logger.info(f"Monitor subprocess started with PID {pid}")
 
-                # Wait for the subprocess to write its PID file
+                # Wait for the subprocess to write its PID file and bind to port
                 # The subprocess will write the PID file after it starts successfully
                 max_wait = 10  # seconds
                 start_time = time.time()
+                pid_file_found = False
+                port_bound = False
+
+                self.logger.debug(f"Waiting up to {max_wait}s for daemon to start...")
 
                 while time.time() - start_time < max_wait:
                     # Check if process is still running
-                    if process.poll() is not None:
-                        # Process exited
+                    returncode = process.poll()
+                    if returncode is not None:
+                        # Process exited - interpret exit code with context
+                        # Exit codes 137 (SIGKILL) and 143 (SIGTERM) are common during daemon replacement
+                        if returncode == EXIT_SIGKILL:
+                            # SIGKILL - process was forcefully terminated
+                            if force_restart:
+                                # This is expected during force restart - old daemon was killed
+                                self.logger.info(
+                                    f"Previous monitor instance replaced (exit {EXIT_SIGKILL}: SIGKILL during force restart)"
+                                )
+                            else:
+                                # Unexpected SIGKILL - something else killed our new daemon
+                                self.logger.warning(
+                                    f"Monitor subprocess terminated unexpectedly (exit {EXIT_SIGKILL}: SIGKILL). "
+                                    f"Check {self.log_file} for details."
+                                )
+                            return False
+                        if returncode == EXIT_SIGTERM:
+                            # SIGTERM - graceful shutdown requested
+                            self.logger.info(
+                                f"Monitor subprocess cleanly terminated (exit {EXIT_SIGTERM}: SIGTERM, graceful shutdown)"
+                            )
+                            return False
+                        if returncode == EXIT_NORMAL:
+                            # Normal exit
+                            self.logger.info(
+                                f"Monitor subprocess exited normally (exit code {EXIT_NORMAL})"
+                            )
+                            return False
+                        # Unexpected exit code - this IS an error
+                        self.logger.error(
+                            f"Monitor daemon subprocess exited prematurely with code {returncode}"
+                        )
                         self.logger.error(
-                            f"Monitor daemon exited with code {process.returncode}"
+                            f"Port {self.port} daemon failed to start. Check {self.log_file} for details."
                         )
                         return False
 
                     # Check if PID file was written
-                    if self.pid_file.exists():
+                    if not pid_file_found and self.pid_file.exists():
                         try:
                             with self.pid_file.open() as f:
                                 written_pid = int(f.read().strip())
                             if written_pid == pid:
-                                # PID file written correctly, check port
-                                if (
-                                    not self._is_port_available()
-                                ):  # Port NOT available means it's in use (good!)
-                                    self.logger.info(
-                                        f"Monitor daemon successfully started on port {self.port}"
-                                    )
-                                    return True
-                        except Exception:
-                            pass  # PID file not ready yet
+                                pid_file_found = True
+                                self.logger.debug(
+                                    f"PID file found with correct PID {pid}"
+                                )
+                        except Exception as e:
+                            self.logger.debug(f"Error reading PID file: {e}")
+
+                    # Check if port is bound (health check)
+                    if not port_bound and not self._is_port_available():
+                        # Port NOT available means it's in use (good!)
+                        port_bound = True
+                        self.logger.debug(f"Port {self.port} is now bound")
+
+                    # Success criteria: both PID file exists and port is bound
+                    if pid_file_found and port_bound:
+                        self.logger.info(
+                            f"Monitor daemon successfully started on port {self.port} (PID: {pid})"
+                        )
+                        # Additional health check: verify we can connect
+                        if self._verify_daemon_health(max_attempts=3):
+                            self.logger.info("Daemon health check passed")
+                            return True
+                        self.logger.warning(
+                            "Daemon started but health check failed - may still be initializing"
+                        )
+                        return (
+                            True  # Return success anyway if PID file and port are good
+                        )
 
                     time.sleep(0.5)