feat: add cold debugging

app.py

@@ -45,28 +45,59 @@ def greet_json():
         "checks": {}
     }
     return health_status
-
+MAX_RETRIES = 3
+RETRY_DELAY = 2  # seconds
 @app.post("/v1/generateTripPlan", response_model=PlanResponse)
 def generate_trip_plan(request: PlanRequest):
     data_importer.coldStartDatabase()
-    try:
-        trip_plan = asyncio.run(agent.query_with_rag(request))
-        return PlanResponse(tripOverview=trip_plan.tripOverview,
-                            query_params=request,
-                            retrieved_data=trip_plan.retrieved_data,
-                            trip_plan=trip_plan.trip_plan,
-                            preparation=trip_plan.preparation,
-                            meta={"status": "success", "timestamp": datetime.utcnow().isoformat()})
-    except Exception as e:
-        logger.error(f"Error in generate_trip_plan: {e}")
-        raise HTTPException(
-            status_code=500,
-            detail={
-                "error": "Internal server error",
-                "message": "Failed to generate trip plan",
-                "details": str(e)
-            }
-        )
+    
+    for attempt in range(MAX_RETRIES):
+        try:
+            logger.info(f"Generating trip plan - attempt {attempt + 1}/{MAX_RETRIES}")
+            trip_plan = asyncio.run(agent.query_with_rag(request))
+            return PlanResponse(
+                tripOverview=trip_plan.tripOverview,
+                query_params=request,
+                retrieved_data=trip_plan.retrieved_data,
+                trip_plan=trip_plan.trip_plan,
+                preparation=trip_plan.preparation,
+                meta={
+                    "status": "success", 
+                    "timestamp": datetime.utcnow().isoformat(),
+                    "attempt": attempt + 1
+                }
+            )
+        except (asyncio.TimeoutError, Timeout, ConnectionError) as timeout_error:
+            logger.warning(f"Timeout error on attempt {attempt + 1}: {timeout_error}")
+            
+            # If this was the last attempt, raise the error
+            if attempt == MAX_RETRIES - 1:
+                logger.error(f"All {MAX_RETRIES} attempts failed due to timeout")
+                raise HTTPException(
+                    status_code=504,  # Gateway Timeout
+                    detail={
+                        "error": "Request timeout",
+                        "message": f"Failed to generate trip plan after {MAX_RETRIES} attempts",
+                        "details": "The service is experiencing high load. Please try again later."
+                    }
+                )
+            
+            # Wait before retrying
+            logger.info(f"Retrying in {RETRY_DELAY} seconds...")
+            time.sleep(RETRY_DELAY)
+            
+        except Exception as e:
+            # For non-timeout errors, don't retry
+            logger.error(f"Non-timeout error in generate_trip_plan: {e}")
+            raise HTTPException(
+                status_code=500,
+                detail={
+                    "error": "Internal server error",
+                    "message": "Failed to generate trip plan",
+                    "details": str(e),
+                    "attempt": attempt + 1
+                }
+            )
 
 
 @app.post("/v1/addYoutubeLink", response_model=YoutubeLinkResponse)