Skip to content

Latest commit

 

History

History
230 lines (154 loc) · 6.63 KB

self-hosting.md

File metadata and controls

230 lines (154 loc) · 6.63 KB
Raw

Hosting your own Shields server

This document describes how to host your own shields server either from source or using a docker image. See the docs on releases for info on how we version the server and how to choose a release.

Installing from Source

You will need Node 12 or later, which you can install using a package manager.

On Ubuntu / Debian:

curl -sL https://deb.nodesource.com/setup_12.x | sudo -E bash -; sudo apt-get install -y nodejs
git clone https://github.com/badges/shields.git
cd shields
git checkout $(git tag | grep server | tail -n 1)  # checkout the latest tag
npm ci  # You may need sudo for this.

Build the frontend

npm run build

Start the server

sudo node server

The server uses port 80 by default, which requires sudo permissions.

There are two ways to provide an alternate port:

PORT=8080 node server
node server 8080

The root gets redirected to https://shields.io.

For testing purposes, you can go to http://localhost/.

Deploying to Heroku

Once you have installed the Heroku CLI

heroku login
heroku create your-app-name
git push heroku master
heroku open

Deploying to Zeit Vercel

To deploy using Zeit Vercel:

npm run build  # Not sure why, but this needs to be run before deploying.
vercel

Docker

DockerHub

We publish images to DockerHub at https://registry.hub.docker.com/r/shieldsio/shields

The next tag is the latest build from master, or tagged releases are available https://registry.hub.docker.com/r/shieldsio/shields/tags

$ docker pull shieldsio/shields:next
$ docker run shieldsio/shields:next

Building Docker Image Locally

Alternatively, you can build and run the server locally using Docker. First build an image:

$ docker build -t shields .
Sending build context to Docker daemon 3.923 MB
…
Successfully built 4471b442c220

Optionally, create a file called shields.env that contains the needed configuration. See server-secrets.md and config/custom-environment-variables.yml for examples.

Then run the container:

$ docker run --rm -p 8080:80 --name shields shields
# or if you have shields.env file, run the following instead
$ docker run --rm -p 8080:80 --env-file shields.env --name shields shields

> badge-maker@3.0.0 start /usr/src/app
> node server.js

http://[::1]/

Assuming Docker is running locally, you should be able to get to the application at http://localhost:8080/.

If you run Docker in a virtual machine (such as boot2docker or Docker Machine) then you will need to replace localhost with the IP address of that virtual machine.

Raster server

If you want to host PNG badges, you can also self-host a raster server which points to your badge server. It's designed as a web function which is tested on Zeit Now, though you may be able to run it on AWS Lambda. It's built on the micro framework, and comes with a start script that allows it to run as a standalone Node service.

  • In your raster instance, set BASE_URL to your Shields instance, e.g. https://shields.example.co.
  • Optionally, in your Shields, instance, configure RASTER_URL to the base URL, e.g. https://raster.example.co. This will send 301 redirects for the legacy raster URLs instead of 404's.

If anyone has set this up, more documentation on how to do this would be welcome! It would also be nice to ship a Docker image that includes a preconfigured raster server.

Persistence

To enable Redis-backed GitHub token persistence, point REDIS_URL to your Redis installation.

Server secrets

You can add your own server secrets in environment variables or config/local.yml.

These are documented in server-secrets.md

Separate frontend hosting

If you want to host the frontend on a separate server, such as cloud storage or a CDN, you can do that.

First, build the frontend, pointing GATSBY_BASE_URL to your server.

GATSBY_BASE_URL=https://your-server.example.com npm run build

Then copy the contents of the build/ folder to your static hosting / CDN.

There are also a couple settings you should configure on the server.

If you want to use server suggestions, you should also set ALLOWED_ORIGIN:

ALLOWED_ORIGIN=http://my-custom-shields.s3.amazonaws.com,https://my-custom-shields.s3.amazonaws.com

This should be a comma-separated list of allowed origin headers. They should not have paths or trailing slashes.

To help out users, you can make the Shields server redirect the server root. Set the REDIRECT_URI environment variable:

REDIRECT_URI=http://my-custom-shields.s3.amazonaws.com/

Sentry

In order to enable integration with Sentry, you need your own Sentry DSN. It’s an URL in format https://{PUBLIC_KEY}:{SECRET_KEY}@sentry.io/{PROJECT_ID}.

How to obtain the Sentry DSN

  1. Sign up for Sentry
  2. Log in to Sentry
  3. Create a new project for Node.js
  4. You should see Sentry DSN for your project. Sentry DSN can be found by navigating to [Project Name] -> Project Settings -> Client Keys (DSN) as well.

Start the server using the Sentry DSN. You can set it:

  • by SENTRY_DSN environment variable
sudo SENTRY_DSN=https://xxx:yyy@sentry.io/zzz node server

Or via config as you would do with server secrets:

private:
  sentry_dsn: ...
sudo node server

Prometheus

Shields uses prom-client to provide default metrics. These metrics are disabled by default. You can enable them by METRICS_PROMETHEUS_ENABLED and METRICS_PROMETHEUS_ENDPOINT_ENABLED environment variables.

METRICS_PROMETHEUS_ENABLED=true METRICS_PROMETHEUS_ENDPOINT_ENABLED=true npm start

Metrics are available at /metrics resource.

Cloudflare

Shields.io uses Cloudflare as a downstream CDN. If your installation does the same, you can configure your server to only accept requests coming from Cloudflare's IPs. Set public.requireCloudflare: true.