File size: 2,337 Bytes
15f1210 20edea9 15f1210 20edea9 e7c1485 20edea9 e7c1485 20edea9 15f1210 e7c1485 15f1210 20edea9 15f1210 e7c1485 20edea9 e7c1485 20edea9 15f1210 20edea9 | 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 | # QModel Docker Compose Configuration — production-safe base
# ============================================================
# This base file is safe to deploy as-is: it does NOT bind-mount the repo,
# so the image's own code is what runs. It ships with just the two data
# files mounted read-only (they're excluded from the built image via
# .dockerignore since they're large binaries that shouldn't bloat it or
# require a rebuild every time the dataset changes).
#
# For local development with live code reload, `docker-compose.override.yml`
# is picked up automatically and adds a full bind-mount on top of this.
#
# Configure via .env file:
# LLM_BACKEND=ollama (default: local Ollama on host machine)
# LLM_BACKEND=hf (HuggingFace backend)
#
# Usage:
# docker-compose up # Dev (uses override.yml automatically)
# docker-compose -f docker-compose.yml up -d # Prod (base file only, no bind-mount)
# docker-compose logs -f # View logs
# docker-compose down # Stop services
services:
qmodel:
build: .
container_name: qmodel-api
ports:
- "8000:8000"
env_file:
- .env
environment:
# Pass through HF token if using HuggingFace backend
- HF_TOKEN=${HF_TOKEN:-}
# Ollama host: override .env to use Docker host IP for container-to-host access
- OLLAMA_HOST=http://host.docker.internal:11434
volumes:
# Data files: read-only, so a data update doesn't require an image rebuild
- ./QModel.index:/app/QModel.index:ro
- ./metadata.json:/app/metadata.json:ro
# Cache HuggingFace models to avoid re-downloading
- huggingface_cache:/root/.cache/huggingface
# Restart automatically if container exits
restart: on-failure:5
extra_hosts:
# Allow container to reach host.docker.internal on Mac/Windows
- "host.docker.internal:host-gateway"
networks:
- qmodel-network
# Health check for orchestration — /health returns 503 until ready
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:8000/health"]
interval: 30s
timeout: 10s
retries: 3
start_period: 60s
networks:
qmodel-network:
driver: bridge
volumes:
# Persistent cache for HuggingFace models
huggingface_cache:
|