Deployment Guide

Deploy DocMirror as a local HTTP service for API integrations and internal evaluation.

Docker Deployment

# Build
docker build -t docmirror:latest .

# Run
docker run -d \
  --name docmirror \
  -p 8000:8000 \
  -v ~/.cache/docmirror:/root/.cache \
  -v ~/.docmirror:/root/.docmirror \
  --restart unless-stopped \
  docmirror:latest

Environment Variables

Variable	Default	Description
DOCMIRROR_API_KEY	—	API key for request authentication
DOCMIRROR_LOG_LEVEL	info	Logging level
DOCMIRROR_MAX_PAGES	200	Maximum pages to process per document
DOCMIRROR_LICENSE	—	Optional commercial license key
OMP_NUM_THREADS	—	Optional native thread limit for OCR/math libraries

Docker Compose

version: "3.8"
services:
  docmirror:
    build: .
    ports:
      - "8000:8000"
    volumes:
      - docmirror_cache:/root/.cache
      - docmirror_data:/root/.docmirror
    environment:
      - DOCMIRROR_API_KEY=${DOCMIRROR_API_KEY}
      - DOCMIRROR_LOG_LEVEL=info
      - OMP_NUM_THREADS=4
    deploy:
      resources:
        limits:
          memory: 2G

volumes:
  docmirror_cache:
  docmirror_data:

Reverse Proxy

Nginx

server {
    listen 443 ssl;
    server_name docmirror.example.com;

    ssl_certificate /etc/ssl/certs/docmirror.crt;
    ssl_certificate_key /etc/ssl/private/docmirror.key;

    client_max_body_size 100M;
    proxy_read_timeout 120s;

    location / {
        proxy_pass http://127.0.0.1:8000;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
}

Caddy

docmirror.example.com {
    reverse_proxy 127.0.0.1:8000
}

Performance Tuning

CPU

• Set OMP_NUM_THREADS to avoid native OCR/math libraries overusing CPU cores.
• For OCR-heavy workloads, start conservatively and raise the thread limit after measuring latency and memory.

Memory

• 2 GB minimum for production
• 4 GB recommended for documents with OCR
• 8 GB for documents with 200+ pages