aion/docker-compose.yml at main · ShubhamDX/aion · 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
services:
  aion:
    build: .
    ports:
      - "8080:8080"
    volumes:
      - ./configs/aion.yaml:/app/configs/aion.yaml:ro
      - aion-data:/app/data
    env_file:
      - path: .env
        required: false
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:8080/health"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 5s
    restart: unless-stopped
    depends_on:
      llama-server:
        condition: service_healthy
        required: false

  # Downloads the GGUF model on first run. Subsequent starts skip the download.
  # The model is stored in a named volume so it persists across restarts.
  model-downloader:
    image: alpine:3.19
    volumes:
      - llama-models:/models
    environment:
      - MODEL_REPO=${LLAMA_MODEL_REPO:-Qwen/Qwen2.5-1.5B-Instruct-GGUF}
      - MODEL_FILE=${LLAMA_MODEL_FILE:-qwen2.5-1.5b-instruct-q4_k_m.gguf}
    entrypoint: /bin/sh
    command:
      - -c
      - |
        TARGET="/models/$${MODEL_FILE}"
        if [ -f "$$TARGET" ]; then
          echo "Model already downloaded: $$TARGET"
          exit 0
        fi
        echo "Downloading $$MODEL_FILE from $$MODEL_REPO ..."
        apk add --no-cache wget > /dev/null 2>&1
        wget -q --show-progress -O "$$TARGET.tmp" \
          "https://huggingface.co/$${MODEL_REPO}/resolve/main/$${MODEL_FILE}" \
          && mv "$$TARGET.tmp" "$$TARGET" \
          && echo "Download complete: $$TARGET" \
          || { echo "Download failed"; rm -f "$$TARGET.tmp"; exit 1; }
    profiles:
      - local

  llama-server:
    image: ghcr.io/ggml-org/llama.cpp:server
    ports:
      - "8081:8081"
    volumes:
      - llama-models:/models:ro
    command: >
      --model /models/${LLAMA_MODEL_FILE:-qwen2.5-1.5b-instruct-q4_k_m.gguf}
      --port 8081
      --host 0.0.0.0
      --ctx-size 4096
      --batch-size 512
    deploy:
      resources:
        limits:
          memory: 8G
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:8081/health"]
      interval: 10s
      timeout: 5s
      retries: 12
      start_period: 60s
    restart: unless-stopped
    depends_on:
      model-downloader:
        condition: service_completed_successfully
    profiles:
      - local

volumes:
  aion-data:
  llama-models: