sharevb/docker-image-download-server repository overview

⁠Docker Image Download Service

A lightweight Node.js HTTP service that lets clients pull, save, and download Docker images on demand.
It supports:

• Multi‑platform pulls (linux/amd64, linux/arm64, etc.)
• Private registry authentication
• Streaming .tar downloads
• Running inside a container with access to the host Docker daemon

This service is ideal for CI pipelines, air‑gapped deployments, offline registries, or tooling that needs to export Docker images programmatically.

⁠Features

⁠✔ Pull Docker images

Uses docker pull with optional --platform.

⁠✔ Save images as tarballs

Uses docker save and streams the result directly to the client.

⁠✔ Registry authentication

Supports:

• Username/password
• Optional registry hostname
• Automatic logout after the request

⁠Requirements

• Docker installed on the host
• Node.js 18+ or 20+
• If running inside Docker:
  mount the host Docker socket

  -v /var/run/docker.sock:/var/run/docker.sock
  

  Copy

⁠Environment Variables

⁠API

⁠GET /download

Downloads a Docker image as a .tar file.

⁠Query parameters

⁠Example: public image

curl -O -J "http://localhost:3000/download?image=alpine:latest"

⁠Example: multi‑arch image

curl -O -J "http://localhost:3000/download?image=nginx:latest&platform=linux/arm64"

⁠Example: private registry

curl -O -J \
  "http://localhost:3000/download?image=myregistry.com/app:1.2.3&platform=linux/amd64&registry=myregistry.com&username=alice&password=secret"

⁠Running Locally

Install dependencies:

npm install

Start the server:

HOST=127.0.0.1 PORT=3000 node server.js

⁠Running in Docker

Pull latest image:

docker pull sharevb/docker-image-download-server:latest

Run it with access to the host Docker daemon:

docker run \
  -p 3000:3000 \
  -v /var/run/docker.sock:/var/run/docker.sock \
  sharevb/docker-image-download-server:latest

⁠Security Notes

• If using registry credentials, prefer short‑lived tokens.
• Use HTTPS or a reverse proxy if exposing the service publicly.

⁠License

MIT