| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | import asyncio |
| 4 | import warnings |
| 5 | |
| 6 | from prometheus_client import start_http_server |
| 7 | |
| 8 | from vllm.logger import init_logger |
| 9 | from vllm.version import __version__ as VLLM_VERSION |
| 10 | |
| 11 | warnings.warn( |
| 12 | "The `python -m vllm.entrypoints.openai.run_batch` command is deprecated " |
| 13 | "and may be removed in a future release. Please use `vllm run_batch` instead.", |
| 14 | DeprecationWarning, |
| 15 | stacklevel=1, |
| 16 | ) |
| 17 | |
| 18 | |
| 19 | if __name__ == "__main__": |
| 20 | from vllm.entrypoints.launchers.run_batch import main, parse_args |
| 21 | |
| 22 | logger = init_logger(__name__) |
| 23 | |
| 24 | args = parse_args() |
| 25 | |
| 26 | logger.info("vLLM batch processing API version %s", VLLM_VERSION) |
| 27 | logger.info("args: %s", args) |
| 28 | |
| 29 | # Start the Prometheus metrics server. LLMEngine uses the Prometheus client |
| 30 | # to publish metrics at the /metrics endpoint. |
| 31 | if args.enable_metrics: |
| 32 | logger.info("Prometheus metrics enabled") |
| 33 | start_http_server(port=args.port, addr=args.host) |
| 34 | else: |
| 35 | logger.info("Prometheus metrics disabled") |
| 36 | |
| 37 | asyncio.run(main(args)) |