frontend.py 2.44 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import subprocess
17
from pathlib import Path
18

19
20
from components.processor import Processor
from components.worker import VllmWorker
21
22
from pydantic import BaseModel

23
from dynamo import sdk
24
25
from dynamo.sdk import depends, service
from dynamo.sdk.lib.config import ServiceConfig
26
from dynamo.sdk.lib.image import DYNAMO_IMAGE
27
28


29
30
31
32
33
34
35
36
37
def get_http_binary_path():
    sdk_path = Path(sdk.__file__)
    binary_path = sdk_path.parent / "cli/bin/http"
    if not binary_path.exists():
        return "http"
    else:
        return str(binary_path)


38
class FrontendConfig(BaseModel):
39
    served_model_name: str
40
41
42
43
44
45
46
    endpoint: str
    port: int = 8080


@service(
    resources={"cpu": "10", "memory": "20Gi"},
    workers=1,
47
    image=DYNAMO_IMAGE,
48
49
50
51
52
53
54
55
56
57
58
)
# todo this should be called ApiServer
class Frontend:
    worker = depends(VllmWorker)
    processor = depends(Processor)

    def __init__(self):
        config = ServiceConfig.get_instance()
        frontend_config = FrontendConfig(**config.get("Frontend", {}))

        subprocess.run(
59
60
61
62
63
64
65
            [
                "llmctl",
                "http",
                "remove",
                "chat-models",
                frontend_config.served_model_name,
            ]
66
67
68
69
70
71
72
        )
        subprocess.run(
            [
                "llmctl",
                "http",
                "add",
                "chat-models",
73
                frontend_config.served_model_name,
74
75
76
77
                frontend_config.endpoint,
            ]
        )

78
        print("Starting HTTP server")
79
        http_binary = get_http_binary_path()
80
        process = subprocess.Popen(
81
            [http_binary, "-p", str(frontend_config.port)], stdout=None, stderr=None
82
        )
83
84
85
86
87
        try:
            process.wait()
        except KeyboardInterrupt:
            process.terminate()
            process.wait()