How can sentry be integrated running vllm
vllm serve /home/ubuntu/models/RedHatAI/Llama-4-Scout-17B-16E-Instruct-quantized.w4a16
–host 0.0.0.0
–port 8000
It seems likely that direct Sentry integration (for error monitoring) is not natively documented in the vLLM serve command or official docs. The available data sources do not mention built-in Sentry support or a specific flag for Sentry in vllm serve or its configuration options. You may need to wrap the vLLM server process with your own Sentry-instrumented Python code or use Sentry’s integrations for FastAPI/Uvicorn, since vLLM’s API server is based on FastAPI/Uvicorn, but this is not explicitly covered in the official vLLM documentation.
Would you like more detail on how to manually add Sentry to a FastAPI/Uvicorn app running vLLM, or guidance on best practices for external monitoring?
Sources:
Hint: Mention @RunLLM in the post for followups.