| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | |
| 4 | |
| 5 | from fastapi import APIRouter, Request |
| 6 | from fastapi.responses import Response |
| 7 | |
| 8 | from vllm.engine.protocol import EngineClient |
| 9 | from vllm.logger import init_logger |
| 10 | from vllm.v1.engine.exceptions import EngineDeadError |
| 11 | |
| 12 | logger = init_logger(__name__) |
| 13 | |
| 14 | |
| 15 | router = APIRouter() |
| 16 | |
| 17 | |
| 18 | def engine_client(request: Request) -> EngineClient: |
| 19 | return request.app.state.engine_client |
| 20 | |
| 21 | |
| 22 | @router.get("/health", response_class=Response) |
| 23 | async def health(raw_request: Request) -> Response: |
| 24 | """Health check.""" |
| 25 | client = engine_client(raw_request) |
| 26 | if client is None: |
| 27 | # Render-only servers have no engine; they are always healthy. |
| 28 | return Response(status_code=200) |
| 29 | try: |
| 30 | await client.check_health() |
| 31 | return Response(status_code=200) |
| 32 | except EngineDeadError: |
| 33 | return Response(status_code=503) |