(parser: FlexibleArgumentParser)
| 227 | |
| 228 | |
| 229 | def make_arg_parser(parser: FlexibleArgumentParser) -> FlexibleArgumentParser: |
| 230 | parser.add_argument("--port", default=8000, type=int, help="port to the http server") |
| 231 | parser.add_argument("--host", default="0.0.0.0", type=str, help="host to the http server") |
| 232 | parser.add_argument("--workers", default=1, type=int, help="number of workers") |
| 233 | parser.add_argument("--metrics-port", default=None, type=int, help="port for metrics server") |
| 234 | parser.add_argument("--controller-port", default=-1, type=int, help="port for controller server") |
| 235 | parser.add_argument( |
| 236 | "--max-waiting-time", |
| 237 | default=-1, |
| 238 | type=int, |
| 239 | help="max waiting time for connection, if set value -1 means no waiting time limit", |
| 240 | ) |
| 241 | parser.add_argument("--max-concurrency", default=512, type=int, help="max concurrency") |
| 242 | |
| 243 | parser.add_argument( |
| 244 | "--enable-mm-output", action="store_true", help="Enable 'multimodal_content' field in response output. " |
| 245 | ) |
| 246 | parser.add_argument( |
| 247 | "--timeout-graceful-shutdown", |
| 248 | default=0, |
| 249 | type=int, |
| 250 | help="timeout for graceful shutdown in seconds (used by gunicorn).Setting it to 0 has the effect of infinite timeouts by disabling timeouts for all workers entirely.", |
| 251 | ) |
| 252 | |
| 253 | parser.add_argument( |
| 254 | "--timeout", |
| 255 | default=0, |
| 256 | type=int, |
| 257 | help="Workers silent for more than this many seconds are killed and restarted.Value is a positive number or 0. Setting it to 0 has the effect of infinite timeouts by disabling timeouts for all workers entirely.", |
| 258 | ) |
| 259 | |
| 260 | parser.add_argument("--api-key", type=str, action="append", help="API_KEY required for service authentication") |
| 261 | |
| 262 | parser = EngineArgs.add_cli_args(parser) |
| 263 | return parser |
| 264 | |
| 265 | |
| 266 | async def listen_for_disconnect(request: Request) -> None: |
no test coverage detected