[serve] Remove unused "endpoint" field of RequestMetadata (#48451)

Some vestigial field that is no longer used...

---------

Signed-off-by: Edward Oakes <ed.nmi.oakes@gmail.com>

java/serve/src/main/java/io/ray/serve/handle/DeploymentHandle.java

@@ -82,7 +82,6 @@ public DeploymentResponse remote(Object... parameters) {
     RayServeMetrics.execute(() -> requestCounter.inc(1.0));
     RequestMetadata.Builder requestMetadata = RequestMetadata.newBuilder();
     requestMetadata.setRequestId(requestContext.getRequestId());
-    requestMetadata.setEndpoint(deploymentId.getName());
     requestMetadata.setCallMethod(
         handleOptions != null ? handleOptions.getMethodName() : Constants.CALL_METHOD);
     requestMetadata.setRoute(requestContext.getRoute());

java/serve/src/test/java/io/ray/serve/util/ServeProtoUtilTest.java

@@ -20,14 +20,12 @@ public void parseRequestMetadataTest() {
         RequestMetadata.newBuilder()
             .setRequestId(requestId)
             .setCallMethod(callMethod)
-            .setEndpoint(endpoint)
             .putContext("context", context)
             .build();
 
     RequestMetadata result = ServeProtoUtil.parseRequestMetadata(requestMetadata.toByteArray());
     Assert.assertNotNull(result);
     Assert.assertEquals(result.getCallMethod(), callMethod);
-    Assert.assertEquals(result.getEndpoint(), endpoint);
     Assert.assertEquals(result.getRequestId(), requestId);
     Assert.assertEquals(result.getContextMap().get("context"), context);
   }

python/ray/serve/_private/common.py

@@ -558,7 +558,6 @@ class RequestMetadata:
     # internal_request_id is always generated by the proxy and is used for tracking
     # request objects. We can assume this is always unique between requests.
     internal_request_id: str
-    endpoint: str
     call_method: str = "__call__"
 
     # HTTP route path of the request.

python/ray/serve/_private/replica.py

@@ -618,7 +618,6 @@ async def handle_request_from_java(
         request_metadata: RequestMetadata = RequestMetadata(
             request_id=proto.request_id,
             internal_request_id=proto.internal_request_id,
-            endpoint=proto.endpoint,
             call_method=proto.call_method,
             multiplexed_model_id=proto.multiplexed_model_id,
             route=proto.route,

python/ray/serve/_private/replica_scheduler/replica_wrapper.py

@@ -119,7 +119,6 @@ def _send_request_java(self, pr: PendingRequest) -> ObjectRef:
         return self._actor_handle.handle_request.remote(
             RequestMetadataProto(
                 request_id=pr.metadata.request_id,
-                endpoint=pr.metadata.endpoint,
                 # Default call method in java is "call," not "__call__" like Python.
                 call_method="call"
                 if pr.metadata.call_method == "__call__"

python/ray/serve/handle.py

@@ -302,7 +302,6 @@ def _remote(
             internal_request_id=_request_context._internal_request_id
             if _request_context._internal_request_id
             else generate_request_id(),
-            endpoint=self.deployment_name,
             call_method=self.handle_options.method_name,
             route=_request_context.route,
             app_name=self.app_name,

python/ray/serve/tests/test_actor_replica_wrapper.py

@@ -116,7 +116,6 @@ async def test_send_request(setup_fake_replica, is_streaming: bool):
         metadata=RequestMetadata(
             request_id="abc",
             internal_request_id="def",
-            endpoint="123",
             is_streaming=is_streaming,
         ),
     )
@@ -149,7 +148,6 @@ async def test_send_request_with_rejection(
         metadata=RequestMetadata(
             request_id="abc",
             internal_request_id="def",
-            endpoint="123",
             is_streaming=is_streaming,
         ),
     )
@@ -187,7 +185,6 @@ async def test_send_request_with_rejection_cancellation(setup_fake_replica):
         metadata=RequestMetadata(
             request_id="abc",
             internal_request_id="def",
-            endpoint="123",
         ),
     )
 

python/ray/serve/tests/unit/test_pow_2_replica_scheduler.py

@@ -187,7 +187,6 @@ def fake_pending_request(
             metadata=RequestMetadata(
                 request_id=str(uuid.uuid4()),
                 internal_request_id=str(uuid.uuid4()),
-                endpoint="endpoint",
                 multiplexed_model_id=model_id,
             ),
             created_at=created_at,
@@ -199,7 +198,6 @@ def fake_pending_request(
             metadata=RequestMetadata(
                 request_id=str(uuid.uuid4()),
                 internal_request_id=str(uuid.uuid4()),
-                endpoint="endpoint",
                 multiplexed_model_id=model_id,
             ),
         )

python/ray/serve/tests/unit/test_proxy.py

@@ -575,7 +575,6 @@ async def test_receive_asgi_messages(self):
         request_metadata = RequestMetadata(
             request_id="fake-request-id",
             internal_request_id="fake-internal-request-id",
-            endpoint="fake-endpoint",
         )
         queue = AsyncMock()
         http_proxy.asgi_receive_queues[internal_request_id] = queue

python/ray/serve/tests/unit/test_request_metadata.py

@@ -14,16 +14,13 @@ def test_request_metadata():
     """
     request_id = "request-id"
     internal_request_id = "internal-request-id"
-    endpoint = "endpoint"
     request_metadata = RequestMetadata(
         request_id=request_id,
         internal_request_id=internal_request_id,
-        endpoint=endpoint,
     )
 
     # Ensure the default values are set correctly.
     assert request_metadata.request_id == request_id
-    assert request_metadata.endpoint == endpoint
     assert request_metadata.call_method == "__call__"
     assert request_metadata.route == ""
     assert request_metadata.app_name == ""

python/ray/serve/tests/unit/test_router.py

@@ -219,7 +219,6 @@ def dummy_request_metadata(is_streaming: bool = False) -> RequestMetadata:
     return RequestMetadata(
         request_id="test-request-1",
         internal_request_id="test-internal-request-1",
-        endpoint="",
         is_streaming=is_streaming,
     )
 
@@ -241,7 +240,6 @@ async def test_basic(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
             is_streaming=is_streaming,
         )
         replica_result = await router.assign_request(request_metadata)
@@ -286,7 +284,6 @@ async def test_basic_with_rejection(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
             is_streaming=is_streaming,
         )
         replica_result = await router.assign_request(request_metadata)
@@ -341,7 +338,6 @@ async def test_retry_with_rejection(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
             is_streaming=is_streaming,
         )
         replica_result = await router.assign_request(request_metadata)
@@ -371,7 +367,6 @@ async def test_cross_lang_no_rejection(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
         )
         replica_result = await router.assign_request(request_metadata)
         assert not replica_result._is_generator_object
@@ -393,7 +388,6 @@ async def test_max_queued_requests_no_limit(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
         )
 
         # Queued a bunch of tasks. None should error because there's no limit.
@@ -431,7 +425,6 @@ async def test_max_queued_requests_limited(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
         )
 
         # Queued `max_queued_requests` tasks. None should fail.
@@ -490,7 +483,6 @@ async def test_max_queued_requests_updated(
         request_metadata = RequestMetadata(
             request_id="test-request-1",
             internal_request_id="test-internal-request-1",
-            endpoint="",
         )
 
         # Queued `max_queued_requests` tasks. None should fail.

python/ray/serve/tests/unit/test_user_callable_wrapper.py

@@ -117,7 +117,6 @@ def _make_request_metadata(
     return RequestMetadata(
         request_id="test_request",
         internal_request_id="test_internal_request",
-        endpoint="test_endpoint",
         call_method=call_method if call_method is not None else "__call__",
         _request_protocol=protocol,
         is_streaming=is_streaming,

src/ray/protobuf/serve.proto

@@ -145,7 +145,7 @@ enum DeploymentLanguage {
 message RequestMetadata {
   string request_id = 1;
 
-  string endpoint = 2;
+  string internal_request_id = 2;
 
   string call_method = 3;
 
@@ -154,8 +154,6 @@ message RequestMetadata {
   string multiplexed_model_id = 5;
 
   string route = 6;
-
-  string internal_request_id = 7;
 }
 
 message RequestWrapper {