hatrack

A simple server for launching mini-apps temporarily.

hatrack will map URL prefixes to launch simple web service commands you configure and proxy requests to them. They'll be terminated after 10 seconds, but restarted if more requests come in to them.

hatrack is a prototype, but has a short and robust time line:

• Configurable timeout with keep-alive as requests come in
• Static file configuration for services
• Logging about worker lifespan
• Recycling ports when the max port number is reached

hatrack is useful if you build lots of mini-projects, want to host them all somewhere, but none of them will get traffic often enough to support running 24/7.

Usage

First, install hatrack with NPM.

npm install -g hatrack

Hatrack will look for a configuration file at ~/.hatrack.yaml is a YAML file like this:

workers:
  test_worker:
    command: "node"
    arguments:
      - "testworker.js"
    prefix: "test"
    port_env: "PORT"
  someproject:
    host: "www.someproject.local"
    cwd: "~/projects/someproject/"
    env:
      PATH: "~/.virtualenvs/someproject/bin/:~/projects/someproject/node_modules/.bin/"
    command: "~/.virtualenvs/someproject/bin/python manage.py runserver $PORT"
  default:
    static: "noservice.html"

workers defines one or more workers and must define a "default" worker.

command is a command to run to launch the worker if it is not working, and will be run with the list of defined arguments. If the command has space-separated arguments, it will be run with a shell and ignores the arguments option.

prefix is the URL prefix that will be routed to this worker. The worker will not see the prefix.

port_env is an environment variable to launch the worker with defining what port number it should listen to.

host defines a hostname to route traffic from to the worker.

env defines one or more environment variables for the command.

cwd defines a current working directory to change to when running the worker.

static defines a file the service should read instead of launching a worker.

timeout defines how long a worker will be kept between usage before being terminated. In ms.

startTimeout defines how long to wait for a worker to come online for the first request. In ms.

Recommended with DNSMasq

The best way to use Hat Rack is combined with DNSMasq. You can configure your machine to route any domain ending in .dev to your localhost, which gives me easy to remember hostnames for all the local projects you run with Hat Rack. You'll be able to open your browser to myproject.dev:8080 and Hat Rack will automatically spin up a development server for myproject to route the request to, and shut it down for you when it is no longer in use.

Install and configure DNSMasq on OSX

First, install DNSMasq via Brew.

brew up
brew install dnsmasq

Once installed, create a simple configuration to listen at your localhost and bind all .dev domains to 127.0.0.1:

cat > /usr/local/etc/dnsmasq.conf
listen-address=127.0.0.1
bind-interfaces
address=/.dev/127.0.0.1

Press Ctrl+C after pasting this command to complete the configuration file.

Now, you'll want to configure DNSMasq to run at start up and to run it immediately:

sudo cp $(brew list dnsmasq | grep /homebrew.mxcl.dnsmasq.plist$) /Library/LaunchDaemons/
sudo launchctl load /Library/LaunchDaemons/homebrew.mxcl.dnsmasq.plist

Now that you've setup DNSMasq to respond to .dev domain resolutions, you need to tell your machine to resolve them through DNSMasq.

sudo mkdir -p /etc/resolver
echo "nameserver 127.0.0.1" > /etc/resolver/dev

Install and configure DNSMasq for Ubuntu

Configure local wildcard DNS server

1. Install Dnsmasq: sudo apt-get install dnsmasq
2. Since Ubuntu's NetworkManager uses dnsmasq, and since that messes things up a little for us, open up /etc/NetworkManager/NetworkManager.conf and comment out (#) the line that reads dns=dnsmasq. Restart NetworkManager afterwards: sudo restart network-manager.
3. Make sure Dnsmasq listens to local DNS queries by editing /etc/dnsmasq.conf, and adding the line listen-address=127.0.0.1.
4. Create a new file in /etc/dnsmasq.d (eg. /etc/dnsmasq.d/dev), and add the line address=/dev/127.0.0.1 to have dnsmasq resolve requests for *.dev domains. Restart Dnsmasq: sudo /etc/init.d/dnsmasq restart.

source: http://brunodbo.be/blog/2013/04/setting-up-wildcard-apache-virtual-host-wildcard-dns

Usage with DNSMasq and .dev domains

Now that you've configured your machine to route all .dev domains locally, you can teach Hat Rack to serve up requests for each of your projects with an entry in the ~/.hatrack.yaml configuration file.

For example, this entry under workers: tells Hat Rack to serve requests to www.ironfroggy.dev by setting some environment variables, cding into my personal website's project directory,

www_dot_ironfroggy_dot_com:
    host: "www.ironfroggy.dev"
    cwd: "~/personal/www-ironfroggy-com/"
    timeout: 15000
    env:
      VIRTUAL_ENV: "/Users/calvin/.virtualenvs/www-ironfroggy-com"
      PATH: "~/.virtualenvs/www-ironfroggy-com/bin/"
    command: "fab reserve:$PORT"

Here we configure:

• host: What .dev domain do we handle with this configuration?
• cwd: What directory does the project exist at?
• timeout: How long do we leave the project running in between requests?
• env: What environment variables are required to run this project?
• command: What command can we use to run the project? This command must use $PORT, which Hat Rack will provide.