Update app.py

app.py

@@ -22,6 +22,7 @@ class LLMManager:
         self.models = {}
         self.models_config = models_config
         self.executor = ThreadPoolExecutor(max_workers=2)
+        self.generation_lock = theading.Lock()
         self.load_all_models()
 
     def load_all_models(self):
@@ -94,30 +95,37 @@ class LLMManager:
 
     def chat_completion(self, model_name, messages, **kwargs):
         """Generar respuesta con modelo específico"""
-        model_data = self.get_model(model_name)
+        if not self.generation_lock.acquire(blocking=False):
+            return {"error": "Servidor ocupado - Generación en progreso"}
+    
+        try:
+            model_data = self.get_model(model_name)
         
-        if not model_data or not model_data["loaded"]:
-            error_msg = f"Modelo {model_name} no cargado"
-            if model_data and "error" in model_data:
-                error_msg += f": {model_data['error']}"
-            return {"error": error_msg}
-
-        response = model_data["instance"].create_chat_completion(
-            messages=messages,
-            **kwargs
-        )
+            if not model_data or not model_data["loaded"]:
+                error_msg = f"Modelo {model_name} no cargado"
+                if model_data and "error" in model_data:
+                    error_msg += f": {model_data['error']}"
+                return {"error": error_msg}
 
-        response["provider"] = "telechars-ai"
-        response["model"] = model_name
-        return response
+            response = model_data["instance"].create_chat_completion(
+                messages=messages,
+                **kwargs
+            )
 
-    def get_loaded_models(self):
-        """Obtener lista de modelos cargados"""
-        loaded = []
-        for name, data in self.models.items():
-            if data["loaded"]:
-                loaded.append(name)
-        return loaded
+            response["provider"] = "telechars-ai"
+            response["model"] = model_name
+            return response
+        
+        finally:
+            # Siempre liberar el lock
+            self.generation_lock.release()
+        def get_loaded_models(self):
+            """Obtener lista de modelos cargados"""
+            loaded = []
+            for name, data in self.models.items():
+                if data["loaded"]:
+                    loaded.append(name)
+            return loaded
 
     def get_all_models_status(self):
         """Obtener estado de todos los modelos"""