🐳 pool 🐳

The simplest proxy service to access your Dockerized web application by Git commit-id, branch or tag.

You can build and run your web application as a Docker container just to access http://<git-commit-id, branch or tag>.pool.dev for example.

---

• Requirements
• Quick start
  • Vagrant
    • Launch Vagrant box
    • Parameters
    • Access web application
  • AWS EC2 instances
    • Instalation for Amazon linux
    • Parameters of init-script
• How it works
• Contributors

Requirements

• Docker should be installed on your host machine
  • Our Vagrantfile is configured to have docker so you don't have to care about it if you try to run on Vagrant
• Web application which you want to preview should be on Git repository
• Git repository of the application has Dockerfile on root of the repository
  • Dockerfile is for a container which the application is going to be run

Quick start

Vagrant

You can run pool on your local Vagrant environment also.

Launch Vagrant box

It needs to install vagrant dns plugin before vagrant up. Just run after vagrant was installed:

vagrant plugin install vagrant-dns
vagrant dns --install
vagrant dns --start

then run:

vagrant up

Parameters

You can give some configuration to pool in Vagrantfile.

Currently we some configurations for pool.

• github-bot
  • It enables github integration
• Git repository url
  • Your Git repository URL
• Maximum numbers of comtainers
  • pool kills containers if the number of containers are over than this number automatically
• Hostname
  • Hostname you want to use to access the environment

You can rewrite Vagrantfile like:

# If you'd like to enable github integration, uncomment below
s.args << "--github-bot"

# Set your repository for previewing by pool
s.args << "https://github.com/mookjp/flaskapp.git"
# Set the maximum number of containers runnning at the same time
s.args << "5"
# Set POOL_BASE_DOMAIN
s.args << [pool_hostname, pool_tld].join(".")

Access web application

To watch the web application on pool, you can do it just to access http://<git-commit-id>.pool.dev with your browser.

In the default setting, pool is configured for the web application maintained in the mookjp/flaskapp repository. You can see the flask app (which just outputs 'hello world') just visiting http://c8f48c60088bbae0d0fb25ed5fd04f4442b58617.pool.dev/ or http://master.pool.dev/.

AWS EC2 instances

Following is an example to install pool to Amazon Linux.

Instalation for Amazon linux

To install pool on your Amazon Linux, use following userdata. This example is for Amazon Linux AMI amzn-ami-hvm-2014.03.2.x86_64-ebs (ami-29dc9228)

#!/bin/sh
# Setup script for pool
# NOTE: Run it as root
yum install -y git
yum install -y docker

# Install latest Docker
service docker stop
curl https://get.docker.com/builds/Linux/x86_64/docker-latest -o /tmp/docker
chmod +x /tmp/docker
cp /tmp/docker /usr/bin/docker
service docker start

# Download pool then run pool container
git clone https://github.com/mookjp/pool.git /app

# You can create `pool` image and run to execute `init_host_server` script.
# It gets 3 parameters:
# 1) Git repository URL
# 2) Maximum number of containers of web application
# 3) Hostname
/app/scripts/init_host_server "https://github.com/mookjp/flaskapp.git" 5 "dev.prevs.io"

Parameters of init-script

pool/scripts/init_host_server is a small util script to run pool container. It gets 3 parameters:

1. Git repository URL
2. Maximum number of containers of web application
3. Hostname. you can get your hostname from your container of application as environment variable; POOL_BASE_DOMAIN so that you can set configuration related to hostname inside your container

Passing a SSH key to pool container

You may want to upload a SSH private key to pool container in case your git repository is hidden from public access.

Then you can place id_rsa at /app/docker/pool/keys/id_rsa in the server where pool is installed.

And id_rsa will be uploaded to /app/keys/id_rsa on the pool container which is going to be used to clone your application repository.

core@pool ~ $ cat<<'EOS'>/app/docker/pool/keys/id_rsa
> -----BEGIN RSA PRIVATE KEY-----
> ..........
> -----END RSA PRIVATE KEY-----
> EOS

core@pool ~ $ sudo /app/scripts/init_host_server

core@pool ~ $ sudo docker exec -it pool bash

[pool container]# ls -l /app/keys/id_rsa

How it works

The part of proxy in pool accesses your Git repository with commit id given as hostname then checkout source with Dockerfile. Dockerfile should be on the root of the repository. After checkout files, the container will be built by the Dockerfile and port is linked with front automatically. All you can do is just to access by URL like http://<git-commit-id, branch or tag>.pool.dev.

pool consists of two modules; proxy hook and container builder. handlers/hook.rb handles HTTP request as proxy. This is a hook script of matsumoto-r/mod_mruby. It forwards port which Docker container was assigned by Git-commit-id.

If there's no container which corresponds to Git-commit-id, build_server works to build Docker image then runs it. build_server sends build log so that you can confirm the status of build process while waiting.

If there is another proccess to build and run container, pool locks to run other process and waits until lock is over.

Contributors

Contributors