[Bug]: V0+TP doesn't work with trust remote code for custom configuration #14925
Description
Your current environment
Nightly
🐛 Describe the bug
vllm --version
INFO 03-17 04:56:52 [__init__.py:256] Automatically detected platform cuda.
0.7.4.dev497+ga73e183e
Use the model such as Deepseek's family of models, which have custom configuration_deepseek.py. In the V0 engine, we see the following error
vllm serve /home/vllm-dev/DeepSeek-V2-Lite --trust-remote-code --tensor-parallel-size 2
INFO 03-17 04:59:17 [__init__.py:256] Automatically detected platform cuda.
INFO 03-17 04:59:18 [api_server.py:972] vLLM API server version 0.7.4.dev497+ga73e183e
INFO 03-17 04:59:18 [api_server.py:973] args: Namespace(subparser='serve', model_tag='/home/vllm-dev/DeepSeek-V2-Lite', config='', host=None, port=8000, uvicorn_log_level='info', allow_credentials=False, allowed_origins=['*'], allowed_methods=['*'], allowed_headers=['*'], api_key=None, lora_modules=None, prompt_adapters=None, chat_template=None, chat_template_content_format='auto', response_role='assistant', ssl_keyfile=None, ssl_certfile=None, ssl_ca_certs=None, enable_ssl_refresh=False, ssl_cert_reqs=0, root_path=None, middleware=[], return_tokens_as_token_ids=False, disable_frontend_multiprocessing=False, enable_request_id_headers=False, enable_auto_tool_choice=False, tool_call_parser=None, tool_parser_plugin='', model='/home/vllm-dev/DeepSeek-V2-Lite', task='auto', tokenizer=None, hf_config_path=None, skip_tokenizer_init=False, revision=None, code_revision=None, tokenizer_revision=None, tokenizer_mode='auto', trust_remote_code=True, allowed_local_media_path=None, download_dir=None, load_format='auto', config_format=<ConfigFormat.AUTO: 'auto'>, dtype='auto', kv_cache_dtype='auto', max_model_len=None, guided_decoding_backend='xgrammar', logits_processor_pattern=None, model_impl='auto', distributed_executor_backend=None, pipeline_parallel_size=1, tensor_parallel_size=2, enable_expert_parallel=False, max_parallel_loading_workers=None, ray_workers_use_nsight=False, block_size=None, enable_prefix_caching=None, disable_sliding_window=False, use_v2_block_manager=True, num_lookahead_slots=0, seed=None, swap_space=4, cpu_offload_gb=0, gpu_memory_utilization=0.9, num_gpu_blocks_override=None, max_num_batched_tokens=None, max_num_partial_prefills=1, max_long_partial_prefills=1, long_prefill_token_threshold=0, max_num_seqs=None, max_logprobs=20, disable_log_stats=False, quantization=None, rope_scaling=None, rope_theta=None, hf_overrides=None, enforce_eager=False, max_seq_len_to_capture=8192, disable_custom_all_reduce=False, tokenizer_pool_size=0, tokenizer_pool_type='ray', tokenizer_pool_extra_config=None, limit_mm_per_prompt=None, mm_processor_kwargs=None, disable_mm_preprocessor_cache=False, enable_lora=False, enable_lora_bias=False, max_loras=1, max_lora_rank=16, lora_extra_vocab_size=256, lora_dtype='auto', long_lora_scaling_factors=None, max_cpu_loras=None, fully_sharded_loras=False, enable_prompt_adapter=False, max_prompt_adapters=1, max_prompt_adapter_token=0, device='auto', num_scheduler_steps=1, use_tqdm_on_load=True, multi_step_stream_outputs=True, scheduler_delay_factor=0.0, enable_chunked_prefill=None, speculative_model=None, speculative_model_quantization=None, num_speculative_tokens=None, speculative_disable_mqa_scorer=False, speculative_draft_tensor_parallel_size=None, speculative_max_model_len=None, speculative_disable_by_batch_size=None, ngram_prompt_lookup_max=None, ngram_prompt_lookup_min=None, spec_decoding_acceptance_method='rejection_sampler', typical_acceptance_sampler_posterior_threshold=None, typical_acceptance_sampler_posterior_alpha=None, disable_logprobs_during_spec_decoding=None, model_loader_extra_config=None, ignore_patterns=[], preemption_mode=None, served_model_name=None, qlora_adapter_name_or_path=None, show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, disable_async_output_proc=False, scheduling_policy='fcfs', scheduler_cls='vllm.core.scheduler.Scheduler', override_neuron_config=None, override_pooler_config=None, compilation_config=None, kv_transfer_config=None, worker_cls='auto', worker_extension_cls='', generation_config='auto', override_generation_config=None, enable_sleep_mode=False, calculate_kv_scales=False, additional_config=None, enable_reasoning=False, reasoning_parser=None, disable_log_requests=False, max_log_len=None, disable_fastapi_docs=False, enable_prompt_tokens_details=False, enable_server_load_tracking=False, dispatch_function=<function ServeSubcommand.cmd at 0x7f0d0102f640>)
INFO 03-17 04:59:18 [config.py:208] Replacing legacy 'type' key with 'rope_type'
INFO 03-17 04:59:24 [config.py:583] This model supports multiple tasks: {'embed', 'classify', 'generate', 'reward', 'score'}. Defaulting to 'generate'.
INFO 03-17 04:59:24 [arg_utils.py:1763] MLA is experimental on VLLM_USE_V1=1. Falling back to V0 Engine.
WARNING 03-17 04:59:24 [arg_utils.py:1639] The model has a long context length (163840). This may causeOOM during the initial memory profiling phase, or result in low performance due to small KV cache size. Consider setting --max-model-len to a smaller value.
INFO 03-17 04:59:24 [config.py:1499] Defaulting to use mp for distributed inference
INFO 03-17 04:59:24 [cuda.py:159] Forcing kv cache block size to 64 for FlashMLA backend.
INFO 03-17 04:59:24 [api_server.py:236] Started engine process with PID 1945303
INFO 03-17 04:59:27 [__init__.py:256] Automatically detected platform cuda.
INFO 03-17 04:59:28 [llm_engine.py:241] Initializing a V0 LLM engine (v0.7.4.dev497+ga73e183e) with config: model='/home/vllm-dev/DeepSeek-V2-Lite', speculative_config=None, tokenizer='/home/vllm-dev/DeepSeek-V2-Lite', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config=None, tokenizer_revision=None, trust_remote_code=True, dtype=torch.bfloat16, max_seq_len=163840, download_dir=None, load_format=auto, tensor_parallel_size=2, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto, device_config=cuda, decoding_config=DecodingConfig(guided_decoding_backend='xgrammar', reasoning_backend=None), observability_config=ObservabilityConfig(show_hidden_metrics=False, otlp_traces_endpoint=None, collect_model_forward_time=False, collect_model_execute_time=False), seed=None, served_model_name=/home/vllm-dev/DeepSeek-V2-Lite, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=None, chunked_prefill_enabled=False, use_async_output_proc=True, disable_mm_preprocessor_cache=False, mm_processor_kwargs=None, pooler_config=None, compilation_config={"splitting_ops":[],"compile_sizes":[],"cudagraph_capture_sizes":[256,248,240,232,224,216,208,200,192,184,176,168,160,152,144,136,128,120,112,104,96,88,80,72,64,56,48,40,32,24,16,8,4,2,1],"max_capture_size":256}, use_cached_outputs=True,
WARNING 03-17 04:59:29 [multiproc_worker_utils.py:310] Reducing Torch parallelism from 64 threads to 1 to avoid unnecessary CPU contention. Set OMP_NUM_THREADS in the external environment to tune this value as needed.
INFO 03-17 04:59:29 [custom_cache_manager.py:19] Setting Triton cache manager to: vllm.triton_utils.custom_cache_manager:CustomCacheManager
ERROR 03-17 04:59:29 [engine.py:443] Can't pickle <class 'transformers_modules.DeepSeek-V2-Lite.configuration_deepseek.DeepseekV2Config'>: it's not the same object as transformers_modules.DeepSeek-V2-Lite.configuration_deepseek.DeepseekV2Config
ERROR 03-17 04:59:29 [engine.py:443] Traceback (most recent call last):
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 431, in run_mp_engine
ERROR 03-17 04:59:29 [engine.py:443] engine = MQLLMEngine.from_vllm_config(
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 126, in from_vllm_config
ERROR 03-17 04:59:29 [engine.py:443] return cls(
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 80, in __init__
ERROR 03-17 04:59:29 [engine.py:443] self.engine = LLMEngine(*args, **kwargs)
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/engine/llm_engine.py", line 280, in __init__
ERROR 03-17 04:59:29 [engine.py:443] self.model_executor = executor_class(vllm_config=vllm_config, )
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/executor/executor_base.py", line 271, in __init__
ERROR 03-17 04:59:29 [engine.py:443] super().__init__(*args, **kwargs)
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/executor/executor_base.py", line 52, in __init__
ERROR 03-17 04:59:29 [engine.py:443] self._init_executor()
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/executor/mp_distributed_executor.py", line 90, in _init_executor
ERROR 03-17 04:59:29 [engine.py:443] worker = ProcessWorkerWrapper(result_handler,
ERROR 03-17 04:59:29 [engine.py:443] File "/home/vllm-dev/simon/bench/vllm/vllm/executor/multiproc_worker_utils.py", line 171, in __init__
ERROR 03-17 04:59:29 [engine.py:443] self.process.start()
ERROR 03-17 04:59:29 [engine.py:443] File "/usr/lib/python3.10/multiprocessing/process.py", line 121, in start
ERROR 03-17 04:59:29 [engine.py:443] self._popen = self._Popen(self)
ERROR 03-17 04:59:29 [engine.py:443] File "/usr/lib/python3.10/multiprocessing/context.py", line 288, in _Popen
ERROR 03-17 04:59:29 [engine.py:443] return Popen(process_obj)
ERROR 03-17 04:59:29 [engine.py:443] File "/usr/lib/python3.10/multiprocessing/popen_spawn_posix.py", line 32, in __init__
ERROR 03-17 04:59:29 [engine.py:443] super().__init__(process_obj)
ERROR 03-17 04:59:29 [engine.py:443] File "/usr/lib/python3.10/multiprocessing/popen_fork.py", line 19, in __init__
ERROR 03-17 04:59:29 [engine.py:443] self._launch(process_obj)
ERROR 03-17 04:59:29 [engine.py:443] File "/usr/lib/python3.10/multiprocessing/popen_spawn_posix.py", line 47, in _launch
ERROR 03-17 04:59:29 [engine.py:443] reduction.dump(process_obj, fp)
ERROR 03-17 04:59:29 [engine.py:443] File "/usr/lib/python3.10/multiprocessing/reduction.py", line 60, in dump
ERROR 03-17 04:59:29 [engine.py:443] ForkingPickler(file, protocol).dump(obj)
ERROR 03-17 04:59:29 [engine.py:443] _pickle.PicklingError: Can't pickle <class 'transformers_modules.DeepSeek-V2-Lite.configuration_deepseek.DeepseekV2Config'>: it's not the same object as transformers_modules.DeepSeek-V2-Lite.configuration_deepseek.DeepseekV2Config
Process SpawnProcess-1:
Traceback (most recent call last):
File "/usr/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
self.run()
File "/usr/lib/python3.10/multiprocessing/process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 445, in run_mp_engine
raise e
File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 431, in run_mp_engine
engine = MQLLMEngine.from_vllm_config(
File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 126, in from_vllm_config
return cls(
File "/home/vllm-dev/simon/bench/vllm/vllm/engine/multiprocessing/engine.py", line 80, in __init__
self.engine = LLMEngine(*args, **kwargs)
File "/home/vllm-dev/simon/bench/vllm/vllm/engine/llm_engine.py", line 280, in __init__
self.model_executor = executor_class(vllm_config=vllm_config, )
File "/home/vllm-dev/simon/bench/vllm/vllm/executor/executor_base.py", line 271, in __init__
super().__init__(*args, **kwargs)
File "/home/vllm-dev/simon/bench/vllm/vllm/executor/executor_base.py", line 52, in __init__
self._init_executor()
File "/home/vllm-dev/simon/bench/vllm/vllm/executor/mp_distributed_executor.py", line 90, in _init_executor
worker = ProcessWorkerWrapper(result_handler,
File "/home/vllm-dev/simon/bench/vllm/vllm/executor/multiproc_worker_utils.py", line 171, in __init__
self.process.start()
File "/usr/lib/python3.10/multiprocessing/process.py", line 121, in start
self._popen = self._Popen(self)
File "/usr/lib/python3.10/multiprocessing/context.py", line 288, in _Popen
return Popen(process_obj)
File "/usr/lib/python3.10/multiprocessing/popen_spawn_posix.py", line 32, in __init__
super().__init__(process_obj)
File "/usr/lib/python3.10/multiprocessing/popen_fork.py", line 19, in __init__
self._launch(process_obj)
File "/usr/lib/python3.10/multiprocessing/popen_spawn_posix.py", line 47, in _launch
reduction.dump(process_obj, fp)
File "/usr/lib/python3.10/multiprocessing/reduction.py", line 60, in dump
ForkingPickler(file, protocol).dump(obj)
_pickle.PicklingError: Can't pickle <class 'transformers_modules.DeepSeek-V2-Lite.configuration_deepseek.DeepseekV2Config'>: it's not the same object as transformers_modules.DeepSeek-V2-Lite.configuration_deepseek.DeepseekV2Config
Traceback (most recent call last):
File "/home/vllm-dev/simon/bench/.venv/bin/vllm", line 10, in <module>
sys.exit(main())
File "/home/vllm-dev/simon/bench/vllm/vllm/entrypoints/cli/main.py", line 75, in main
args.dispatch_function(args)
File "/home/vllm-dev/simon/bench/vllm/vllm/entrypoints/cli/serve.py", line 33, in cmd
uvloop.run(run_server(args))
File "/home/vllm-dev/simon/bench/.venv/lib/python3.10/site-packages/uvloop/__init__.py", line 82, in run
return loop.run_until_complete(wrapper())
File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
File "/home/vllm-dev/simon/bench/.venv/lib/python3.10/site-packages/uvloop/__init__.py", line 61, in wrapper
return await main
File "/home/vllm-dev/simon/bench/vllm/vllm/entrypoints/openai/api_server.py", line 1007, in run_server
async with build_async_engine_client(args) as engine_client:
File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
return await anext(self.gen)
File "/home/vllm-dev/simon/bench/vllm/vllm/entrypoints/openai/api_server.py", line 139, in build_async_engine_client
async with build_async_engine_client_from_engine_args(
File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
return await anext(self.gen)
File "/home/vllm-dev/simon/bench/vllm/vllm/entrypoints/openai/api_server.py", line 259, in build_async_engine_client_from_engine_args
raise RuntimeError(
RuntimeError: Engine process failed to start. See stack trace for the root cause.
Before submitting a new issue...
- Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.