local-opencode-stack/docker-compose.yml at main · P2Enjoy/local-opencode-stack · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
services:
  vllm-node:
    build:
      context: .
      dockerfile: Dockerfile
    container_name: vllm-container
    # restart: unless-stopped

    ports:
      - "8000:8000"

    gpus: all

    env_file:
      - ./vars.env
    environment:
      # MODEL is selected by runMe.sh via `env MODEL=<name> docker compose up`.
      # Keep a default for direct `docker compose up` usage.
      MODEL: ${MODEL:-glm47-flash}
      # VARIANT selects the config variant within a family model file (e.g. nvfp4).
      # Leave empty for legacy single-model files.
      VARIANT: ${VARIANT:-}

    volumes:
      - ${HOME}/.cache/huggingface:/root/.cache/huggingface
      - ./scripts/run_vllm_agent.sh:/app/run_vllm_agent.sh:ro
      - ./models:/app/models:ro
      - ./secrets:/app/secrets:ro
      - ./scripts:/app/scripts:ro
      - ./vars.env:/app/vars.env:ro
      - ./scripts/generate_litellm_config.py:/app/generate_litellm_config.py:ro
      - ./scripts/litellm_config.template.yaml:/app/litellm_config.template.yaml:ro
      - litellm_config:/app/generated_configs
      - ./moe_configs/${MODEL:-glm47-flash}:/usr/local/lib/python3.12/dist-packages/vllm/model_executor/layers/fused_moe/configs:ro

    command: bash /app/run_vllm_agent.sh

    healthcheck:
      test: ["CMD-SHELL", "test -f /app/generated_configs/config.yaml && curl -s -H 'Authorization: Bearer sk-FAKE' http://localhost:8000/v1/models | grep -q vllm_agent || exit 1"]
      interval: 15s
      timeout: 60s
      retries: 50
      start_period: 150s

  litellm:
    image: ghcr.io/berriai/litellm:main-stable
    command: ["--config=/app/generated_configs/config.yaml"]
    container_name: litellm
    # restart: unless-stopped

    volumes:
      - litellm_config:/app/generated_configs

    ports:
      - "4000:4000"

    env_file:
      - ./vars.env

    depends_on:
      db:
        condition: service_healthy
      vllm-node:
        condition: service_healthy

    healthcheck:
      test: ["CMD-SHELL", "wget --no-verbose --tries=1 http://localhost:4000/health/liveliness || exit 1"]
      interval: 30s
      timeout: 30s
      retries: 10
      start_period: 30s

  db:
    image: postgres:16
    restart: always
    container_name: litellm_db

    environment:
      POSTGRES_DB: litellm
      POSTGRES_USER: llmproxy
      POSTGRES_PASSWORD: dbpassword9090

    volumes:
      - postgres_data:/var/lib/postgresql/data

    healthcheck:
      test: ["CMD-SHELL", "pg_isready -d litellm -U llmproxy"]
      interval: 1s
      timeout: 5s
      retries: 10

volumes:
  postgres_data:
    name: litellm_postgres_data
  litellm_config:
    name: litellm_config
    driver: local
    driver_opts:
      type: tmpfs
      device: tmpfs