Skip to content

adding asyncenginedead error catch #504

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ian-scale merged 3 commits into from
Apr 30, 2024
Merged

adding asyncenginedead error catch #504

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
30ab834
adding asyncenginedead error catch
ian-scale-2 Apr 29, 2024
97f4ce5
catch error in generation
ian-scale-2 Apr 30, 2024
b0c005c
Merge branch 'main' into ianmacleod/add_async_error_handling
ian-scale Apr 30, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 7 additions & 2 deletions model-engine/model_engine_server/inference/vllm/vllm_server.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@
from fastapi import BackgroundTasks, FastAPI, HTTPException, Request
from fastapi.responses import Response, StreamingResponse
from vllm.engine.arg_utils import AsyncEngineArgs
from vllm.engine.async_llm_engine import AsyncLLMEngine
from vllm.engine.async_llm_engine import AsyncEngineDeadError, AsyncLLMEngine
from vllm.entrypoints.openai.protocol import CompletionRequest as OpenAICompletionRequest
from vllm.model_executor.guided_decoding import get_guided_decoding_logits_processor
from vllm.outputs import CompletionOutput
Expand Down Expand Up @@ -75,7 +75,11 @@ async def generate(request: Request) -> Response:
sampling_params.logits_processors.append(guided_decode_logit_processor)

request_id = random_uuid()
results_generator = engine.generate(prompt, sampling_params, request_id)
try:
results_generator = engine.generate(prompt, sampling_params, request_id)
except AsyncEngineDeadError as e:
print(f"The vllm engine is dead, exiting the pod: {e}")
exit(1)

# Streaming case
async def stream_results() -> AsyncGenerator[str, None]:
Expand Down Expand Up @@ -192,6 +196,7 @@ def extract_logprobs(logprobs: Dict[int, Logprob]) -> Dict[int, float]:

engine_args = AsyncEngineArgs.from_cli_args(args)
engine = AsyncLLMEngine.from_engine_args(engine_args)
engine.check_health()

signal.signal(signal.SIGUSR1, debug)

Expand Down