name: gpu-lab-sglang
services:
  sglang:
    image: gpu-lab-sglang:v100-cu121
    container_name: gpu-lab-sglang
    network_mode: host
    ipc: host
    shm_size: "32gb"
    env_file:
      - active/current.env
    environment:
      - PYTHONPATH=/opt/v100-bootstrap
    volumes:
      - /root/models:/models:ro
      - /root/model-cache:/model-cache:ro
      - /root/gpu-lab/start-sglang-qwen.sh:/start-sglang-qwen.sh:ro
      - /root/gpu-lab/v100-bootstrap:/opt/v100-bootstrap:ro
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    entrypoint: ["/bin/bash", "/start-sglang-qwen.sh"]
    restart: "no"