Deprecate --disable-log-requests and replace with --enable-log-requests (#21739)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

.buildkite/nightly-benchmarks/README.md

@@ -104,7 +104,6 @@ We test the throughput by using `vllm bench serve` with request rate = inf to co
             "tensor_parallel_size": 1,
             "swap_space": 16,
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "load_format": "dummy"
         },
         "client_parameters": {

.buildkite/nightly-benchmarks/tests/genai-perf-tests.json

@@ -11,7 +11,6 @@
         },
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,

.buildkite/nightly-benchmarks/tests/nightly-tests.json

@@ -35,7 +35,6 @@
         }, 
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,
@@ -90,7 +89,6 @@
         }, 
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,
@@ -145,7 +143,6 @@
         }, 
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,
@@ -197,7 +194,6 @@
         }, 
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,
@@ -251,7 +247,6 @@
         }, 
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,
@@ -305,7 +300,6 @@
         }, 
         "vllm_server_parameters": {
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "gpu_memory_utilization": 0.9,
             "num_scheduler_steps": 10,
             "max_num_seqs": 512,

.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json

@@ -17,7 +17,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -50,7 +49,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -83,7 +81,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -117,7 +114,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -153,7 +149,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -189,7 +184,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,

.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json

@@ -17,7 +17,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -50,7 +49,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -84,7 +82,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -118,7 +115,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -154,7 +150,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -191,7 +186,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,

.buildkite/nightly-benchmarks/tests/serving-tests-cpu.json

@@ -17,7 +17,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -50,7 +49,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -83,7 +81,6 @@
 	    "block_size": 128,
 	    "trust_remote_code": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -117,7 +114,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,
@@ -153,7 +149,6 @@
 	    "trust_remote_code": "",
 	    "enable_chunked_prefill": "",
             "disable_log_stats": "",
-            "disable_log_requests": "",
 	    "enforce_eager": "",
 	    "max_num_batched_tokens": 2048,
 	    "max_num_seqs": 256,

.buildkite/nightly-benchmarks/tests/serving-tests.json

@@ -7,7 +7,6 @@
             "tensor_parallel_size": 1,
             "swap_space": 16,
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "load_format": "dummy"
         },
         "client_parameters": {
@@ -26,7 +25,6 @@
             "tensor_parallel_size": 4,
             "swap_space": 16,
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "load_format": "dummy"
         },
         "client_parameters": {
@@ -45,7 +43,6 @@
             "tensor_parallel_size": 2,
             "swap_space": 16,
             "disable_log_stats": "",
-            "disable_log_requests": "",
             "load_format": "dummy"
         },
         "client_parameters": {
@@ -60,8 +57,7 @@
         "test_name": "serving_llama70B_tp4_sharegpt_specdecode",
         "qps_list": [2],
         "server_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
+            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct", 
-            "disable_log_requests": "", 
             "tensor_parallel_size": 4,
             "swap_space": 16,
             "speculative_config": {

tests/config/test_mp_reducer.py

@@ -28,7 +28,6 @@ def test_mp_reducer(monkeypatch):
             max_model_len=32,
             gpu_memory_utilization=0.1,
             disable_log_stats=True,
-            disable_log_requests=True,
         )
 
         async_llm = AsyncLLM.from_engine_args(

tests/mq_llm_engine/test_load.py

@@ -16,7 +16,7 @@ NUM_EXPECTED_TOKENS = 10
 NUM_REQUESTS = 10000
 
 # Scenarios to test for num generated token.
-ENGINE_ARGS = AsyncEngineArgs(model=MODEL, disable_log_requests=True)
+ENGINE_ARGS = AsyncEngineArgs(model=MODEL)
 
 
 @pytest.fixture(scope="function")

tests/v1/engine/test_async_llm.py

@@ -26,12 +26,10 @@ if not current_platform.is_cuda():
 TEXT_ENGINE_ARGS = AsyncEngineArgs(
     model="meta-llama/Llama-3.2-1B-Instruct",
     enforce_eager=True,
-    disable_log_requests=True,
 )
 
 VISION_ENGINE_ARGS = AsyncEngineArgs(model="Qwen/Qwen2-VL-2B-Instruct",
-                                     enforce_eager=True,
+                                     enforce_eager=True)
-                                     disable_log_requests=True)
 
 TEXT_PROMPT = "Hello my name is Robert and"
 

tests/v1/test_async_llm_dp.py

@@ -25,7 +25,6 @@ DP_SIZE = int(os.getenv("DP_SIZE", 2))
 engine_args = AsyncEngineArgs(
     model="ibm-research/PowerMoE-3b",
     enforce_eager=True,
-    disable_log_requests=True,
     tensor_parallel_size=int(os.getenv("TP_SIZE", 1)),
     data_parallel_size=DP_SIZE,
 )

vllm/engine/arg_utils.py

@@ -18,7 +18,7 @@ from typing import (TYPE_CHECKING, Annotated, Any, Callable, Dict, List,
 import regex as re
 import torch
 from pydantic import TypeAdapter, ValidationError
-from typing_extensions import TypeIs
+from typing_extensions import TypeIs, deprecated
 
 import vllm.envs as envs
 from vllm.config import (BlockSize, CacheConfig, CacheDType, CompilationConfig,
@@ -1704,7 +1704,23 @@ class EngineArgs:
 @dataclass
 class AsyncEngineArgs(EngineArgs):
     """Arguments for asynchronous vLLM engine."""
-    disable_log_requests: bool = False
+    enable_log_requests: bool = False
+
+    @property
+    @deprecated(
+        "`disable_log_requests` is deprecated and has been replaced with "
+        "`enable_log_requests`. This will be removed in v0.12.0. Please use "
+        "`enable_log_requests` instead.")
+    def disable_log_requests(self) -> bool:
+        return not self.enable_log_requests
+
+    @disable_log_requests.setter
+    @deprecated(
+        "`disable_log_requests` is deprecated and has been replaced with "
+        "`enable_log_requests`. This will be removed in v0.12.0. Please use "
+        "`enable_log_requests` instead.")
+    def disable_log_requests(self, value: bool):
+        self.enable_log_requests = not value
 
     @staticmethod
     def add_cli_args(parser: FlexibleArgumentParser,
@@ -1715,9 +1731,15 @@ class AsyncEngineArgs(EngineArgs):
         load_general_plugins()
         if not async_args_only:
             parser = EngineArgs.add_cli_args(parser)
+        parser.add_argument('--enable-log-requests',
+                            action=argparse.BooleanOptionalAction,
+                            default=AsyncEngineArgs.enable_log_requests,
+                            help='Enable logging requests.')
         parser.add_argument('--disable-log-requests',
-                            action='store_true',
+                            action=argparse.BooleanOptionalAction,
-                            help='Disable logging requests.')
+                            default=not AsyncEngineArgs.enable_log_requests,
+                            help='[DEPRECATED] Disable logging requests.',
+                            deprecated=True)
         current_platform.pre_register_and_update(parser)
         return parser
 

vllm/engine/async_llm_engine.py

@@ -30,7 +30,7 @@ from vllm.sampling_params import SamplingParams
 from vllm.sequence import ExecuteModelRequest
 from vllm.transformers_utils.tokenizer import AnyTokenizer
 from vllm.usage.usage_lib import UsageContext
-from vllm.utils import Device, weak_bind
+from vllm.utils import Device, deprecate_kwargs, weak_bind
 
 logger = init_logger(__name__)
 ENGINE_ITERATION_TIMEOUT_S = envs.VLLM_ENGINE_ITERATION_TIMEOUT_S
@@ -554,14 +554,20 @@ class AsyncLLMEngine(EngineClient):
         return LLMEngine._get_executor_cls(engine_config)
 
     @classmethod
+    @deprecate_kwargs(
+        "disable_log_requests",
+        additional_message=("This argument will have no effect. "
+                            "Use `enable_log_requests` instead."),
+    )
     def from_vllm_config(
-        cls,
+            cls,
-        vllm_config: VllmConfig,
+            vllm_config: VllmConfig,
-        start_engine_loop: bool = True,
+            start_engine_loop: bool = True,
-        usage_context: UsageContext = UsageContext.ENGINE_CONTEXT,
+            usage_context: UsageContext = UsageContext.ENGINE_CONTEXT,
-        stat_loggers: Optional[dict[str, StatLoggerBase]] = None,
+            stat_loggers: Optional[dict[str, StatLoggerBase]] = None,
-        disable_log_requests: bool = False,
+            enable_log_requests: bool = False,
-        disable_log_stats: bool = False,
+            disable_log_stats: bool = False,
+            disable_log_requests: bool = True,  # Deprecated, will be removed
     ) -> "AsyncLLMEngine":
         """Create an AsyncLLMEngine from the EngineArgs."""
 
@@ -569,7 +575,7 @@ class AsyncLLMEngine(EngineClient):
             vllm_config=vllm_config,
             executor_class=cls._get_executor_cls(vllm_config),
             start_engine_loop=start_engine_loop,
-            log_requests=not disable_log_requests,
+            log_requests=enable_log_requests,
             log_stats=not disable_log_stats,
             usage_context=usage_context,
             stat_loggers=stat_loggers,
@@ -598,7 +604,7 @@ class AsyncLLMEngine(EngineClient):
             usage_context=usage_context,
             stat_loggers=stat_loggers,
             disable_log_stats=engine_args.disable_log_stats,
-            disable_log_requests=engine_args.disable_log_requests,
+            enable_log_requests=engine_args.enable_log_requests,
         )
 
     @property

vllm/engine/multiprocessing/engine.py

@@ -34,6 +34,7 @@ from vllm.outputs import RequestOutput
 from vllm.transformers_utils.config import (
     maybe_register_config_serialize_by_value)
 from vllm.usage.usage_lib import UsageContext
+from vllm.utils import deprecate_kwargs
 from vllm.worker.model_runner_base import InputProcessingError
 
 logger = init_logger(__name__)
@@ -120,10 +121,20 @@ class MQLLMEngine:
             return ENGINE_DEAD_ERROR()
 
     @classmethod
-    def from_vllm_config(cls, vllm_config: VllmConfig,
+    @deprecate_kwargs(
-                         usage_context: UsageContext,
+        "disable_log_requests",
-                         disable_log_requests: bool, disable_log_stats: bool,
+        additional_message=("This argument will have no effect. "
-                         ipc_path: str) -> "MQLLMEngine":
+                            "Use `enable_log_requests` instead."),
+    )
+    def from_vllm_config(
+            cls,
+            vllm_config: VllmConfig,
+            usage_context: UsageContext,
+            enable_log_requests: bool,
+            disable_log_stats: bool,
+            ipc_path: str,
+            disable_log_requests: bool = True,  # Deprecated, will be removed
+    ) -> "MQLLMEngine":
         # Setup plugins for each process
         from vllm.plugins import load_general_plugins
         load_general_plugins()
@@ -136,7 +147,7 @@ class MQLLMEngine:
             ipc_path=ipc_path,
             usage_context=usage_context,
             use_async_sockets=use_async_sockets,
-            log_requests=(not disable_log_requests),
+            log_requests=enable_log_requests,
             log_stats=(not disable_log_stats),
         )
 
@@ -150,7 +161,7 @@ class MQLLMEngine:
             ipc_path=ipc_path,
             vllm_config=vllm_config,
             usage_context=usage_context,
-            disable_log_requests=engine_args.disable_log_requests,
+            enable_log_requests=engine_args.enable_log_requests,
             disable_log_stats=engine_args.disable_log_stats,
         )
 
@@ -436,7 +447,7 @@ def signal_handler(*_) -> None:
 
 def run_mp_engine(vllm_config: VllmConfig, usage_context: UsageContext,
                   ipc_path: str, disable_log_stats: bool,
-                  disable_log_requests: bool, engine_alive):
+                  enable_log_requests: bool, engine_alive):
     try:
         # Ensure we can serialize transformer config before spawning
         maybe_register_config_serialize_by_value()
@@ -445,7 +456,7 @@ def run_mp_engine(vllm_config: VllmConfig, usage_context: UsageContext,
             vllm_config=vllm_config,
             usage_context=usage_context,
             disable_log_stats=disable_log_stats,
-            disable_log_requests=disable_log_requests,
+            enable_log_requests=enable_log_requests,
             ipc_path=ipc_path)
 
         signal.signal(signal.SIGTERM, signal_handler)

vllm/entrypoints/openai/api_server.py

@@ -205,7 +205,7 @@ async def build_async_engine_client_from_engine_args(
             async_llm = AsyncLLM.from_vllm_config(
                 vllm_config=vllm_config,
                 usage_context=usage_context,
-                disable_log_requests=engine_args.disable_log_requests,
+                enable_log_requests=engine_args.enable_log_requests,
                 disable_log_stats=engine_args.disable_log_stats,
                 client_addresses=client_config,
                 client_index=client_index)
@@ -227,7 +227,7 @@ async def build_async_engine_client_from_engine_args(
             engine_client = AsyncLLMEngine.from_vllm_config(
                 vllm_config=vllm_config,
                 usage_context=usage_context,
-                disable_log_requests=engine_args.disable_log_requests,
+                enable_log_requests=engine_args.enable_log_requests,
                 disable_log_stats=engine_args.disable_log_stats)
             yield engine_client
         finally:
@@ -272,7 +272,7 @@ async def build_async_engine_client_from_engine_args(
             target=run_mp_engine,
             args=(vllm_config, UsageContext.OPENAI_API_SERVER, ipc_path,
                   engine_args.disable_log_stats,
-                  engine_args.disable_log_requests, engine_alive))
+                  engine_args.enable_log_requests, engine_alive))
         engine_process.start()
         engine_pid = engine_process.pid
         assert engine_pid is not None, "Engine process failed to start."
@@ -1570,10 +1570,10 @@ async def init_app_state(
     else:
         served_model_names = [args.model]
 
-    if args.disable_log_requests:
+    if args.enable_log_requests:
-        request_logger = None
-    else:
         request_logger = RequestLogger(max_log_len=args.max_log_len)
+    else:
+        request_logger = None
 
     base_model_paths = [
         BaseModelPath(name=name, model_path=args.model)

vllm/entrypoints/openai/run_batch.py

@@ -324,10 +324,10 @@ async def run_batch(
     else:
         served_model_names = [args.model]
 
-    if args.disable_log_requests:
+    if args.enable_log_requests:
-        request_logger = None
-    else:
         request_logger = RequestLogger(max_log_len=args.max_log_len)
+    else:
+        request_logger = None
 
     base_model_paths = [
         BaseModelPath(name=name, model_path=args.model)

vllm/utils/__init__.py

@@ -1668,6 +1668,12 @@ class FlexibleArgumentParser(ArgumentParser):
         # Enable the deprecated kwarg for Python 3.12 and below
 
         def parse_known_args(self, args=None, namespace=None):
+            if args is not None and "--disable-log-requests" in args:
+                # Special case warning because the warning below won't trigger
+                # if –-disable-log-requests because its value is default.
+                logger.warning_once(
+                    "argument '--disable-log-requests' is deprecated. This "
+                    "will be removed in v0.12.0.")
             namespace, args = super().parse_known_args(args, namespace)
             for action in FlexibleArgumentParser._deprecated:
                 if (hasattr(namespace, dest := action.dest)

vllm/v1/engine/async_llm.py

@@ -27,7 +27,7 @@ from vllm.transformers_utils.config import (
 from vllm.transformers_utils.tokenizer import AnyTokenizer
 from vllm.transformers_utils.tokenizer_group import init_tokenizer_from_configs
 from vllm.usage.usage_lib import UsageContext
-from vllm.utils import Device, cdiv
+from vllm.utils import Device, cdiv, deprecate_kwargs
 from vllm.v1.engine import EngineCoreRequest
 from vllm.v1.engine.core_client import EngineCoreClient
 from vllm.v1.engine.exceptions import EngineDeadError, EngineGenerateError
@@ -142,16 +142,22 @@ class AsyncLLM(EngineClient):
             pass
 
     @classmethod
+    @deprecate_kwargs(
+        "disable_log_requests",
+        additional_message=("This argument will have no effect. "
+                            "Use `enable_log_requests` instead."),
+    )
     def from_vllm_config(
-        cls,
+            cls,
-        vllm_config: VllmConfig,
+            vllm_config: VllmConfig,
-        start_engine_loop: bool = True,
+            start_engine_loop: bool = True,
-        usage_context: UsageContext = UsageContext.ENGINE_CONTEXT,
+            usage_context: UsageContext = UsageContext.ENGINE_CONTEXT,
-        stat_loggers: Optional[list[StatLoggerFactory]] = None,
+            stat_loggers: Optional[list[StatLoggerFactory]] = None,
-        disable_log_requests: bool = False,
+            enable_log_requests: bool = False,
-        disable_log_stats: bool = False,
+            disable_log_stats: bool = False,
-        client_addresses: Optional[dict[str, str]] = None,
+            client_addresses: Optional[dict[str, str]] = None,
-        client_index: int = 0,
+            client_index: int = 0,
+            disable_log_requests: bool = True,  # Deprecated, will be removed
     ) -> "AsyncLLM":
         if not envs.VLLM_USE_V1:
             raise ValueError(
@@ -166,7 +172,7 @@ class AsyncLLM(EngineClient):
             executor_class=Executor.get_class(vllm_config),
             start_engine_loop=start_engine_loop,
             stat_loggers=stat_loggers,
-            log_requests=not disable_log_requests,
+            log_requests=enable_log_requests,
             log_stats=not disable_log_stats,
             usage_context=usage_context,
             client_addresses=client_addresses,
@@ -191,7 +197,7 @@ class AsyncLLM(EngineClient):
         return cls(
             vllm_config=vllm_config,
             executor_class=executor_class,
-            log_requests=not engine_args.disable_log_requests,
+            log_requests=engine_args.enable_log_requests,
             log_stats=not engine_args.disable_log_stats,
             start_engine_loop=start_engine_loop,
             usage_context=usage_context,