Implement async lifespan for FastAPI and refactor HTTP client usage i…

…n surface query service

backend_py/primary/primary/main.py

@@ -57,7 +57,30 @@ def custom_generate_unique_id(route: APIRoute) -> str:
     return f"{route.name}"
 
 
+
+
+
+import httpx
+from contextlib import asynccontextmanager
+
+@asynccontextmanager
+async def app_lifespan(app: FastAPI):
+    print("start lifespan")
+    #limits = httpx.Limits(max_keepalive_connections=200, max_connections=300)
+
+    #app.state.requests_client = httpx.AsyncClient(http2=True, verify=False, limits=limits)
+    #app.state.requests_client = httpx.AsyncClient(verify=False)
+    app.state.requests_client = httpx.AsyncClient()
+    yield
+
+    await app.state.requests_client.aclose()
+    print("end lifespan")
+
+
+
+
 app = FastAPI(
+    lifespan=app_lifespan,
     generate_unique_id_function=custom_generate_unique_id,
     root_path="/api",
     default_response_class=ORJSONResponse,

backend_py/primary/primary/routers/surface/router.py

@@ -2,7 +2,7 @@
 import logging
 from typing import Annotated, List, Optional, Literal
 
-from fastapi import APIRouter, Depends, HTTPException, Query, Response, Body, status
+from fastapi import APIRouter, Depends, HTTPException, Query, Response, Request, Body, status
 from webviz_pkg.core_utils.perf_metrics import PerfMetrics
 
 from primary.services.sumo_access.case_inspector import CaseInspector
@@ -231,6 +231,7 @@ async def post_get_surface_intersection(
 
 @router.post("/sample_surface_in_points")
 async def post_sample_surface_in_points(
+    request: Request,
     case_uuid: str = Query(description="Sumo case uuid"),
     ensemble_name: str = Query(description="Ensemble name"),
     surface_name: str = Query(description="Surface name"),
@@ -242,7 +243,10 @@ async def post_sample_surface_in_points(
 
     sumo_access_token = authenticated_user.get_sumo_access_token()
 
+    async_client = request.app.state.requests_client
+
     result_arr: List[RealizationSampleResult] = await batch_sample_surface_in_points_async(
+        async_client=async_client,
         sumo_access_token=sumo_access_token,
         case_uuid=case_uuid,
         iteration_name=ensemble_name,

backend_py/primary/primary/services/surface_query_service/surface_query_service.py

@@ -48,6 +48,7 @@ class _PointSamplingResponseBody(BaseModel):
 
 
 async def batch_sample_surface_in_points_async(
+    async_client: httpx.AsyncClient,
     sumo_access_token: str,
     case_uuid: str,
     iteration_name: str,
@@ -117,14 +118,19 @@ async def batch_sample_surface_in_points_async(
 
     json_request_body = request_body.model_dump()
 
-    async with httpx.AsyncClient(timeout=300) as client:
-        LOGGER.info(f"Running async go point sampling for surface: {surface_name}")
+    LOGGER.info(f"Running async go point sampling for surface: {surface_name}")
+    perf_metrics.record_lap("prepare_call")
 
-        perf_metrics.record_lap("prepare_call")
 
-        response: httpx.Response = await client.post(
-            url=SERVICE_ENDPOINT, json=json_request_body
-        )
+    response: httpx.Response = await async_client.post(url=SERVICE_ENDPOINT, json=json_request_body)
+
+
+    # async with httpx.AsyncClient(timeout=300) as client:
+    #     LOGGER.info(f"Running async go point sampling for surface: {surface_name}")
+
+    #     perf_metrics.record_lap("prepare_call")
+
+    #     response: httpx.Response = await client.post(url=SERVICE_ENDPOINT, json=json_request_body)
 
     perf_metrics.record_lap("main-call")