posters-science-extraction-api/docker-compose.dev.yml at main · fairdataihub/posters-science-extraction-api · 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
# Development configuration for docker-compose
#
# Usage:
#   First time (builds image, downloads models ~16GB each):
#     docker compose -f docker-compose.dev.yml up --build
#
#   After editing Python files (fast restart, no rebuild):
#     docker compose -f docker-compose.dev.yml restart
#
#   View logs:
#     docker compose -f docker-compose.dev.yml logs -f
#
#   If you change the model (JSON_MODEL_ID or VISION_MODEL_ID), clear the cache:
#     docker compose -f docker-compose.dev.yml down -v
#     docker compose -f docker-compose.dev.yml up --build

services:
  poster-extraction:
    build:
      context: .
      dockerfile: Dockerfile
    image: poster-extraction:dev
    container_name: poster-extraction-dev
    env_file:
      - .env
    environment:
      - CUDA_VISIBLE_DEVICES=${CUDA_VISIBLE_DEVICES:-0}
      - NVIDIA_VISIBLE_DEVICES=all
      - NVIDIA_DRIVER_CAPABILITIES=compute,utility
      - PORT=8000
      - HOST=0.0.0.0
      - DATABASE_URL=${DATABASE_URL:-}
      - BUNNY_PRIVATE_STORAGE=${BUNNY_PRIVATE_STORAGE:-}
      - BUNNY_PRIVATE_STORAGE_KEY=${BUNNY_PRIVATE_STORAGE_KEY:-}
      - BUNNY_PUBLIC_STORAGE=${BUNNY_PUBLIC_STORAGE:-}
      - BUNNY_PUBLIC_STORAGE_KEY=${BUNNY_PUBLIC_STORAGE_KEY:-}
      - POLL_INTERVAL_SECONDS=${POLL_INTERVAL_SECONDS:-30}
      - STUCK_PROCESSING_MINUTES=${STUCK_PROCESSING_MINUTES:-5}
    ports:
      - "${API_PORT:-8000}:8000"
    # Mount source code for hot reload - edit locally, restart to apply
    volumes:
      - ./poster_extraction.py:/app/poster_extraction.py
      - ./api.py:/app/api.py
      - ./validation.py:/app/validation.py
      - ./poster_extraction_schema.json:/app/poster_extraction_schema.json
      - ./poster_schema.json:/app/poster_schema.json
      # Persist model cache between container rebuilds
      - model-cache:/root/.cache
    stdin_open: true
    tty: true
    restart: "no"
    deploy:
      resources:
        limits:
          memory: 32G
        reservations:
          memory: 16G
          devices:
            - driver: nvidia
              count: ${GPU_COUNT:-1}
              capabilities: [gpu]

# Named volume for model cache persistence
volumes:
  model-cache: