[Refactor] fix ruff rule B039: mutable-contextvar-default (#49854)

## Why are these changes needed?

Replace mutable default value with None to prevent shared state across
ContextVar.get() calls

ref : https://docs.astral.sh/ruff/rules/mutable-contextvar-default/

## Related issue number

#47991 

---------

Signed-off-by: LeoLiao123 <[email protected]>

pyproject.toml

@@ -38,7 +38,6 @@ ignore = [
     "B026",
     "B027",
     "B035",
-    "B039",
     "B904",
     "C416",
     "C419",

python/ray/serve/_private/default_impl.py

@@ -93,7 +93,7 @@ def create_init_handle_options(**kwargs):
 
 
 def get_request_metadata(init_options, handle_options):
-    _request_context = ray.serve.context._serve_request_context.get()
+    _request_context = ray.serve.context._get_serve_request_context()
 
     request_protocol = RequestProtocol.UNDEFINED
     if init_options and init_options._source == DeploymentHandleSource.PROXY:

python/ray/serve/_private/logging_utils.py

@@ -84,7 +84,7 @@ class ServeContextFilter(logging.Filter):
     """
 
     def filter(self, record):
-        request_context = ray.serve.context._serve_request_context.get()
+        request_context = ray.serve.context._get_serve_request_context()
         if request_context.route:
             setattr(record, SERVE_LOG_ROUTE, request_context.route)
         if request_context.request_id:

python/ray/serve/_private/proxy.py

@@ -456,7 +456,7 @@ async def proxy_request(self, proxy_request: ProxyRequest) -> ResponseGenerator:
 
         latency_ms = (time.time() - start_time) * 1000.0
         if response_handler_info.should_record_access_log:
-            request_context = ray.serve.context._serve_request_context.get()
+            request_context = ray.serve.context._get_serve_request_context()
             logger.info(
                 access_log_msg(
                     method=proxy_request.method,

python/ray/serve/_private/router.py

@@ -620,7 +620,7 @@ async def assign_request(
 
                 # Keep track of requests that have been sent out to replicas
                 if RAY_SERVE_COLLECT_AUTOSCALING_METRICS_ON_HANDLE:
-                    _request_context = ray.serve.context._serve_request_context.get()
+                    _request_context = ray.serve.context._get_serve_request_context()
                     request_id: str = _request_context.request_id
                     self._metrics_manager.inc_num_running_requests_for_replica(
                         replica_id

python/ray/serve/api.py

@@ -725,7 +725,7 @@ def get_multiplexed_model_id() -> str:
             def my_deployment_function(request):
                 assert serve.get_multiplexed_model_id() == "model_1"
     """
-    _request_context = ray.serve.context._serve_request_context.get()
+    _request_context = ray.serve.context._get_serve_request_context()
     return _request_context.multiplexed_model_id
 
 

python/ray/serve/context.py

@@ -181,10 +181,22 @@ class _RequestContext:
 
 
 _serve_request_context = contextvars.ContextVar(
-    "Serve internal request context variable", default=_RequestContext()
+    "Serve internal request context variable", default=None
 )
 
 
+def _get_serve_request_context():
+    """Get the current request context.
+
+    Returns:
+        The current request context
+    """
+
+    if _serve_request_context.get() is None:
+        _serve_request_context.set(_RequestContext())
+    return _serve_request_context.get()
+
+
 def _set_request_context(
     route: str = "",
     request_id: str = "",
@@ -195,7 +207,8 @@ def _set_request_context(
     """Set the request context. If the value is not set,
     the current context value will be used."""
 
-    current_request_context = _serve_request_context.get()
+    current_request_context = _get_serve_request_context()
+
     _serve_request_context.set(
         _RequestContext(
             route=route or current_request_context.route,

python/ray/serve/metrics.py

@@ -60,7 +60,7 @@ def _add_serve_metric_default_tags(default_tags: Dict[str, str]):
 def _add_serve_context_tag_values(tag_keys: Tuple, tags: Dict[str, str]):
     """Add serve context tag values to the metric tags"""
 
-    _request_context = ray.serve.context._serve_request_context.get()
+    _request_context = ray.serve.context._get_serve_request_context()
     if ROUTE_TAG in tag_keys and ROUTE_TAG not in tags:
         tags[ROUTE_TAG] = _request_context.route
 

python/ray/serve/tests/test_http_headers.py

@@ -20,7 +20,7 @@ def test_request_id_header_by_default(serve_instance):
     @serve.deployment
     class Model:
         def __call__(self):
-            request_id = ray.serve.context._serve_request_context.get().request_id
+            request_id = ray.serve.context._get_serve_request_context().request_id
             return request_id
 
     serve.run(Model.bind())
@@ -52,7 +52,7 @@ def test_basic(self, serve_instance):
         @serve.deployment
         class Model:
             def __call__(self) -> int:
-                request_id = ray.serve.context._serve_request_context.get().request_id
+                request_id = ray.serve.context._get_serve_request_context().request_id
                 assert request_id == "123-234"
                 return 1
 
@@ -67,7 +67,7 @@ def test_fastapi(self, serve_instance):
         class Model:
             @app.get("/")
             def say_hi(self) -> int:
-                request_id = ray.serve.context._serve_request_context.get().request_id
+                request_id = ray.serve.context._get_serve_request_context().request_id
                 assert request_id == "123-234"
                 return 1
 
@@ -78,7 +78,7 @@ def test_starlette_resp(self, serve_instance):
         @serve.deployment
         class Model:
             def __call__(self) -> int:
-                request_id = ray.serve.context._serve_request_context.get().request_id
+                request_id = ray.serve.context._get_serve_request_context().request_id
                 assert request_id == "123-234"
                 return starlette.responses.Response("1", media_type="application/json")
 
@@ -94,7 +94,7 @@ def test_set_request_id_headers_with_two_attributes(serve_instance):
     @serve.deployment
     class Model:
         def __call__(self):
-            request_id = ray.serve.context._serve_request_context.get().request_id
+            request_id = ray.serve.context._get_serve_request_context().request_id
             return request_id
 
     serve.run(Model.bind())
@@ -128,7 +128,7 @@ def test_reuse_request_id(serve_instance):
     class MyFastAPIDeployment:
         @app.post("/hello")
         def root(self, user_input: Dict[str, str]) -> Dict[str, str]:
-            request_id = ray.serve.context._serve_request_context.get().request_id
+            request_id = ray.serve.context._get_serve_request_context().request_id
             return {
                 "app_name": user_input["app_name"],
                 "serve_context_request_id": request_id,

python/ray/serve/tests/test_logging.py

@@ -341,7 +341,7 @@ def test_context_information_in_logging(serve_and_ray_shutdown, json_log_format)
     )
     def fn(*args):
         logger.info("user func")
-        request_context = ray.serve.context._serve_request_context.get()
+        request_context = ray.serve.context._get_serve_request_context()
         return {
             "request_id": request_context.request_id,
             "route": request_context.route,
@@ -362,7 +362,7 @@ def fn(*args):
     class Model:
         def __call__(self, req: starlette.requests.Request):
             logger.info("user log message from class method")
-            request_context = ray.serve.context._serve_request_context.get()
+            request_context = ray.serve.context._get_serve_request_context()
             return {
                 "request_id": request_context.request_id,
                 "route": request_context.route,

python/ray/serve/tests/test_replica_request_context.py

@@ -6,11 +6,11 @@
 from fastapi.responses import PlainTextResponse
 
 from ray import serve
-from ray.serve.context import _serve_request_context
+from ray.serve.context import _get_serve_request_context
 
 
 def _get_request_context_route() -> str:
-    return _serve_request_context.get().route
+    return _get_serve_request_context().route
 
 
 class TestHTTPRoute:

python/ray/serve/tests/test_replica_sync_methods.py

@@ -83,7 +83,7 @@ class D:
             @fastapi_app.get("/")
             def root(self):
                 return PlainTextResponse(
-                    serve.context._serve_request_context.get().request_id
+                    serve.context._get_serve_request_context().request_id
                 )
 
     else:
@@ -92,7 +92,7 @@ def root(self):
         class D:
             def __call__(self) -> str:
                 return PlainTextResponse(
-                    serve.context._serve_request_context.get().request_id
+                    serve.context._get_serve_request_context().request_id
                 )
 
     serve.run(D.bind())