docker-compose.oldgpu.yml

version: "3.4"

services:
  traefik:
    image: "traefik:v2.10"
    container_name: "traefik"
    command:
      - "--log.level=DEBUG"
      - "--api.insecure=true"
      - "--providers.docker=true"
      - "--providers.docker.exposedbydefault=false"
      - "--entrypoints.web.address=:8070"
      - "--accesslog"
      - "--api"
    ports:
      - "8070:8070"
      #- "8080:8080"
    volumes:
      - "/var/run/docker.sock:/var/run/docker.sock:ro"

  whoami:
    # A container that exposes an API to show its IP address
    image: traefik/whoami
    labels:
      - traefik.http.routers.whoami.rule=Host(`whoami.localhost`)
      - traefik.http.routers.whoami.entrypoints=web
      - traefik.enable=true

  whisper-asr-webservice-1:
    image: allomediadocker/openai-whisper-asr-webservice:cuda-10.2
    environment:
      # https://github.com/openai/whisper#available-models-and-languages
      - ASR_MODEL=large
      - ASR_ENGINE=openai_whisper # faster_whisper can't work with cuda10 (only cuda 11)
    env_file:
      - .env
    deploy:
      mode: replicated
      replicas: 1
      resources:
        reservations:
          devices:
          - driver: nvidia
            capabilities: [gpu]
            device_ids: ['0']
    labels:
      - traefik.enable=true
      - traefik.http.routers.whisper-asr-webservice.rule=Host(`${TRAEFIK_HOSTNAME}`)
      - traefik.http.routers.whisper-asr-webservice.entrypoints=web
      - traefik.http.services.whisper-asr-webservice.loadbalancer.server.port=9000
    volumes:
      - ./app:/app/app
      - cache-pip:/root/.cache/pip
      - cache-poetry:/root/.cache/poetry
      - cache-whisper:/root/.cache/whisper

  whisper-asr-webservice-2:
    image: allomediadocker/openai-whisper-asr-webservice:cuda-10.2
    environment:
      # https://github.com/openai/whisper#available-models-and-languages
      - ASR_MODEL=large
      - ASR_ENGINE=openai_whisper
    env_file:
      - .env
    deploy:
      mode: replicated
      replicas: 1
      resources:
        reservations:
          devices:
          - driver: nvidia
            capabilities: [gpu]
            device_ids: ['1']
    labels:
      - traefik.enable=true
      - traefik.http.routers.whisper-asr-webservice.rule=Host(`${TRAEFIK_HOSTNAME}`)
      - traefik.http.routers.whisper-asr-webservice.entrypoints=web
      - traefik.http.services.whisper-asr-webservice.loadbalancer.server.port=9000
    volumes:
      - ./app:/app/app
      - cache-pip:/root/.cache/pip
      - cache-poetry:/root/.cache/poetry
      - cache-whisper:/root/.cache/whisper

  whisper-asr-webservice-3:
    image: allomediadocker/openai-whisper-asr-webservice:cuda-10.2
    environment:
      # https://github.com/openai/whisper#available-models-and-languages
      - ASR_MODEL=large
      - ASR_ENGINE=openai_whisper
    env_file:
      - .env
    deploy:
      mode: replicated
      replicas: 1
      resources:
        reservations:
          devices:
          - driver: nvidia
            capabilities: [gpu]
            device_ids: ['2']
    labels:
      - traefik.enable=true
      - traefik.http.routers.whisper-asr-webservice.rule=Host(`${TRAEFIK_HOSTNAME}`)
      - traefik.http.routers.whisper-asr-webservice.entrypoints=web
      - traefik.http.services.whisper-asr-webservice.loadbalancer.server.port=9000
    volumes:
      - ./app:/app/app
      - cache-pip:/root/.cache/pip
      - cache-poetry:/root/.cache/poetry
      - cache-whisper:/root/.cache/whisper

volumes:
  cache-pip:
  cache-poetry:
  cache-whisper:
  cache-faster-whisper: