Do not throw ResourceQuotaExceeded

qdrant_client/common/client_exceptions.py

@@ -14,11 +14,3 @@ def __init__(self, message: str, retry_after_s: int) -> None:
 
     def __str__(self) -> str:
         return self.message.strip()
-
-
-class ResourceQuotaExceeded(QdrantException):
-    def __init__(self, message: str) -> None:
-        self.message = message if message else "Quota Exceeded Response"
-
-    def __str__(self) -> str:
-        return self.message.strip()

qdrant_client/connection.py

@@ -4,7 +4,7 @@
 
 import grpc
 
-from qdrant_client.common.client_exceptions import ResourceExhaustedResponse, ResourceQuotaExceeded
+from qdrant_client.common.client_exceptions import ResourceExhaustedResponse
 
 
 # type: ignore # noqa: F401
@@ -145,8 +145,6 @@ def process_response(response: Any) -> Any:
             reason_phrase = response.details() if response.details() else ""
             if retry_after:
                 raise ResourceExhaustedResponse(message=reason_phrase, retry_after_s=retry_after)
-            else:
-                raise ResourceQuotaExceeded(message=reason_phrase)
         return response
 
     def intercept_call(

qdrant_client/http/api_client.py

@@ -5,7 +5,7 @@
 
 from httpx import AsyncClient, Client, Request, Response
 from pydantic import ValidationError
-from qdrant_client.common.client_exceptions import ResourceExhaustedResponse, ResourceQuotaExceeded
+from qdrant_client.common.client_exceptions import ResourceExhaustedResponse
 from qdrant_client.http.api.aliases_api import AsyncAliasesApi, SyncAliasesApi
 from qdrant_client.http.api.beta_api import AsyncBetaApi, SyncBetaApi
 from qdrant_client.http.api.collections_api import AsyncCollectionsApi, SyncCollectionsApi
@@ -119,8 +119,6 @@ def send(self, request: Request, type_: Type[T]) -> T:
 
             if retry_after_s:
                 raise ResourceExhaustedResponse(message, retry_after_s)
-            else:
-                raise ResourceQuotaExceeded(message=message)
 
         if response.status_code in [200, 201, 202]:
             try:
@@ -208,8 +206,6 @@ async def send(self, request: Request, type_: Type[T]) -> T:
 
             if retry_after_s:
                 raise ResourceExhaustedResponse(message, retry_after_s)
-            else:
-                raise ResourceQuotaExceeded(message=message)
 
         if response.status_code in [200, 201, 202]:
             try:

tests/congruence_tests/test_rate_limit.py

@@ -2,9 +2,11 @@
 
 import numpy as np
 import pytest
+from grpc import RpcError
 
-from qdrant_client.common.client_exceptions import ResourceQuotaExceeded, ResourceExhaustedResponse
+from qdrant_client.common.client_exceptions import ResourceExhaustedResponse
 from qdrant_client.http import models
+from qdrant_client.http.exceptions import UnexpectedResponse
 from tests.congruence_tests.test_common import (
     COLLECTION_NAME,
     compare_collections,
@@ -45,13 +47,13 @@ def test_upsert_hits_large_request_limit(remote_client):
     )
 
     with pytest.raises(
-        ResourceQuotaExceeded,
+        UnexpectedResponse,
         match="Write rate limit exceeded, request larger than than rate limiter capacity, please try to split your request",
     ):
         remote_client.upsert(COLLECTION_NAME, points_batch)
 
     with pytest.raises(
-        ResourceQuotaExceeded,
+        RpcError,
         match="Write rate limit exceeded, request larger than than rate limiter capacity, please try to split your request",
     ):
         grpc_client.upsert(COLLECTION_NAME, points_batch)
@@ -123,8 +125,39 @@ def test_upsert_hits_write_rate_limit(remote_client):
 def test_upload_collection_succeeds_with_limits(local_client, remote_client):
     grpc_client = init_remote(prefer_grpc=True)
 
-    points = generate_fixtures(10)
+    remote_client.update_collection(
+        collection_name=COLLECTION_NAME,
+        strict_mode_config=models.StrictModeConfig(
+            enabled=True, read_rate_limit=READ_LIMIT, write_rate_limit=WRITE_LIMIT
+        ),
+    )
+
+    # pre-condition: hit the limit first then do upload_collection
+    points = generate_fixtures(WRITE_LIMIT)
+    ids, payload = [], []
+    vectors = {}
+    for point in points:
+        ids.append(point.id)
+        payload.append(point.payload)
+        for vector_name, vector in point.vector.items():
+            if vector_name not in vectors:
+                vectors[vector_name] = []
+            vectors[vector_name].append(vector)
 
+    points_batch = models.Batch(
+        ids=ids,
+        vectors=vectors,
+        payloads=payload,
+    )
+    try:
+        for _ in range(10):
+            remote_client.upsert(COLLECTION_NAME, points_batch)
+            grpc_client.upsert(COLLECTION_NAME, points_batch)
+    except ResourceExhaustedResponse as ex:
+        print(ex.message)
+    # end of pre-condition
+
+    points = generate_fixtures(WRITE_LIMIT)
     vectors = []
     payload = []
     ids = []
@@ -133,13 +166,6 @@ def test_upload_collection_succeeds_with_limits(local_client, remote_client):
         vectors.append(point.vector)
         payload.append(point.payload)
 
-    remote_client.update_collection(
-        collection_name=COLLECTION_NAME,
-        strict_mode_config=models.StrictModeConfig(
-            enabled=True, read_rate_limit=READ_LIMIT, write_rate_limit=WRITE_LIMIT
-        ),
-    )
-
     local_client.upload_collection(COLLECTION_NAME, vectors, payload, ids=ids)
     remote_client.upload_collection(
         COLLECTION_NAME, vectors, payload, ids=ids, wait=True, max_retries=1
@@ -152,25 +178,6 @@ def test_upload_collection_succeeds_with_limits(local_client, remote_client):
     compare_collections(local_client, grpc_client, UPLOAD_NUM_VECTORS)
 
 
-def test_upload_points_succeeds_with_limits(local_client, remote_client):
-    grpc_client = init_remote(prefer_grpc=True)
-    points = generate_fixtures(10)
-
-    remote_client.update_collection(
-        collection_name=COLLECTION_NAME,
-        strict_mode_config=models.StrictModeConfig(
-            enabled=True, read_rate_limit=READ_LIMIT, write_rate_limit=WRITE_LIMIT
-        ),
-    )
-
-    local_client.upload_points(COLLECTION_NAME, points)
-    remote_client.upload_points(COLLECTION_NAME, points, wait=True, max_retries=1)
-    grpc_client.upload_points(COLLECTION_NAME, points, wait=True, max_retries=1)
-
-    compare_collections(local_client, remote_client, UPLOAD_NUM_VECTORS)
-    compare_collections(local_client, grpc_client, UPLOAD_NUM_VECTORS)
-
-
 def test_query_hits_read_rate_limit(remote_client):
     grpc_client = init_remote(prefer_grpc=True)