cloudflare-docker-proxy

If you're looking for proxy for helm, maybe you can try cloudflare-helm-proxy.

Deploy

1. click the "Deploy With Workers" button
2. follow the instructions to fork and deploy
3. update routes as you requirement

Routes configuration tutorial

1. use cloudflare worker host: only support proxy one registry

   const routes = {
     "${workername}.${username}.workers.dev/": "https://registry-1.docker.io",
   };

2. use custom domain: support proxy multiple registries route by host

   • host your domain DNS on cloudflare
   • add A record of xxx.example.com to 192.0.2.1
   • deploy this project to cloudflare workers
   • add xxx.example.com/* to HTTP routes of workers
   • add more records and modify the config as you need

   const routes = {
     "docker.libcuda.so": "https://registry-1.docker.io",
     "quay.libcuda.so": "https://quay.io",
     "gcr.libcuda.so": "https://k8s.gcr.io",
     "k8s-gcr.libcuda.so": "https://k8s.gcr.io",
     "ghcr.libcuda.so": "https://ghcr.io",
   };

3. deploy by wrangler

# replace libcuda.so to your domain in wrangler.toml
sed -i 's/libcuda.so/${CUSTOM_DOMAIN}/g' wrangler.toml
# install install dependencies
yarn
# deploy
yarn wrangler deploy --env production --minify src/index.js

Usage

# busybox:stable ==> docker.libcuda.so/busybox:stable
docker pull docker.libcuda.so/busybox:stable
# or
sudo mkdir -p /etc/docker
sudo tee /etc/docker/daemon.json <<-EOF
{
    "registry-mirrors": [
        "https://docker.libcuda.so",
    ]
}
EOF
sudo systemctl daemon-reload
sudo systemctl restart docker
docker pull busybox:stable