xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-07-09 10:07:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Add type assertion of request_id for LLMEngine.add_request (#19700)

Browse Source 
Signed-off-by: n2ptr <xuzhanchaomail@163.com>
This commit is contained in:
Chales Xu 2025-06-28 13:47:36 +08:00 committed by GitHub
parent c329ceca6d
commit e53be6f00a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 14 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
tests/mq_llm_engine/test_error_handling.py
View File

@ -66,7 +66,7 @@ async def test_evil_forward(tmp_socket):
with pytest.raises(MQEngineDeadError): with pytest.raises(MQEngineDeadError):
async for _ in client.generate(prompt="Hello my name is", async for _ in client.generate(prompt="Hello my name is",
sampling_params=SamplingParams(), sampling_params=SamplingParams(),
request_id=uuid.uuid4()): request_id=str(uuid.uuid4())):
pass pass
assert client.errored assert client.errored
@ -115,7 +115,7 @@ async def test_failed_health_check(tmp_socket):
with pytest.raises(MQEngineDeadError): with pytest.raises(MQEngineDeadError):
async for _ in client.generate(prompt="Hello my name is", async for _ in client.generate(prompt="Hello my name is",
sampling_params=SamplingParams(), sampling_params=SamplingParams(),
request_id=uuid.uuid4()): request_id=str(uuid.uuid4())):
pass pass
client.close() client.close()
@ -157,7 +157,7 @@ async def test_failed_abort(tmp_socket):
async for _ in client.generate( async for _ in client.generate(
prompt="Hello my name is", prompt="Hello my name is",
sampling_params=SamplingParams(max_tokens=10), sampling_params=SamplingParams(max_tokens=10),
request_id=uuid.uuid4()): request_id=str(uuid.uuid4())):
pass pass
assert "KeyError" in repr(execinfo.value) assert "KeyError" in repr(execinfo.value)
assert client.errored assert client.errored
@ -189,7 +189,7 @@ async def test_batch_error(tmp_socket):
params = SamplingParams(min_tokens=2048, max_tokens=2048) params = SamplingParams(min_tokens=2048, max_tokens=2048)
async for _ in client.generate(prompt="Hello my name is", async for _ in client.generate(prompt="Hello my name is",
sampling_params=params, sampling_params=params,
request_id=uuid.uuid4()): request_id=str(uuid.uuid4())):
pass pass
tasks = [asyncio.create_task(do_generate(client)) for _ in range(10)] tasks = [asyncio.create_task(do_generate(client)) for _ in range(10)]
@ -289,7 +289,7 @@ async def test_engine_process_death(tmp_socket):
with pytest.raises(MQEngineDeadError): with pytest.raises(MQEngineDeadError):
async for _ in client.generate(prompt="Hello my name is", async for _ in client.generate(prompt="Hello my name is",
sampling_params=SamplingParams(), sampling_params=SamplingParams(),
request_id=uuid.uuid4()): request_id=str(uuid.uuid4())):
pass pass
# And the health check should show the engine is dead # And the health check should show the engine is dead

4
vllm/engine/llm_engine.py
View File

@ -687,6 +687,10 @@ class LLMEngine:
>>> # continue the request processing >>> # continue the request processing
>>> ... >>> ...
""" """
if not isinstance(request_id, str):
raise TypeError(
f"request_id must be a string, got {type(request_id)}")
if lora_request is not None and not self.lora_config: if lora_request is not None and not self.lora_config:
raise ValueError(f"Got lora_request {lora_request} but LoRA is " raise ValueError(f"Got lora_request {lora_request} but LoRA is "
"not enabled!") "not enabled!")

5
vllm/v1/engine/llm_engine.py
View File

@ -192,6 +192,11 @@ class LLMEngine:
prompt_adapter_request: Optional[PromptAdapterRequest] = None, prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0, priority: int = 0,
) -> None: ) -> None:
# Validate the request_id type.
if not isinstance(request_id, str):
raise TypeError(
f"request_id must be a string, got {type(request_id)}")
# Process raw inputs into the request. # Process raw inputs into the request.
prompt_str, request = self.processor.process_inputs( prompt_str, request = self.processor.process_inputs(
request_id, prompt, params, arrival_time, lora_request, request_id, prompt, params, arrival_time, lora_request,