The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
 / 
App-Padadoy-0.125
River stage zero No dependents
++
/ App::Padadoy

NAME

App::Padadoy - Simply deploy PSGI applications

VERSION

version 0.125

SYNOPSIS

Create a new application and start it locally on your development machine:

  $ padadoy create Your::Module
  $ plackup app/app.psgi

Start application locally as deamon with bundled dependencies:

  $ padadoy cartontest
  $ padadoy start

Show status of your running application and stop it:

  $ padadoy status
  $ padadoy stop

Manage your application files in a git repository:

  $ git add *
  $ git commit -m "inial commit"

Deploy the application at dotCloud

  $ dotcloud create nameoryourapp
  $ dotcloud push nameofyourapp

Prepare your own deployment machine (as remote in padadoy.yml):

  $ padadoy remote init

Add your deployment machine as git remote and deploy:

  $ git remote add prod ...
  $ git push prod master

DESCRIPTION

This is an early preview release, be warned! Design changes are likely, at least until a stable carton 1.0 has been released!

Padadoy is a command line application to facilitate deployment of PSGI applications, inspired by http://dotcloud.com. Padadoy is based on the Carton module dependency manager, Starman webserver, and git. In short, an application is managed in a git repository and pushed to a remote repository for deployment. At the remote server, required modules are installed and unit tests are run, to minimize the chance of a broken installation.

An application is managed in a git repository with following structure. You can create it automatically with padadoy create or padadoy create Your::App::Module.

    app/
       app.psgi      - application startup script
       lib/          - local perl modules (at least the actual application)
       t/            - unit tests
       Makefile.PL   - used to determine required modules and to run tests

    deplist.txt      - a list of perl modules required to run (o)
      
    data/            - persistent data (o)

    dotcloud.yml     - basic configuration for dotCloud (o)
    
    libs -> app/lib                - symlink for OpenShift (o)
    deplist.txt -> app/deplist.txt - symlink for OpenShift (o)
    perl/index.pl                  - CGI script for OpenShift (o)

This directory layout helps to easy deploy on multiple platforms. Files and directories marked by (o) are optional, depending on what platform you want to deploy. Padadoy also facilitates deploying to your own servers just like a PaaS provider.

On the deployment machine there is a directory with the following structure:

    repository/      - the bare git repository that the app is pushed to
    current -> ...   - symbolic link to the current working directory
    new -> ...       - symbolic link to the new working directory on updates
    padadoy.yml      - local configuration

You can create this layout with padadoy remote init. After adding the remote repository as git remote, you can simply deploy new versions with git push.

METHODS

new ( [$configfile] [%configvalues] )

Start padadoy, optionally with some configuration (padadoy.yml).

create

Create an application boilerplate.

deplist

List dependencies (not implemented yet).

init

Initialize on your deployment machine.

config

Show configuration values.

restart

Start or gracefully restart the application if running.

start

Start starman webserver with carton.

stop

Stop starman webserver.

status

Show some status information.

checkout ( [$revision], [$directory], [$current] )

Check out a revision to a new working directory. If no directory name is specified, the revision name will be concatenated to the base directory. If a current directory is specified, the local directory will first be copied with rsync to avoid reinstallation of dependent packages with carton.

cartontest

Update dependencies with carton and run tests.

update ( [$revision] )

Checkout a revision, test it, and create a symlink called new on success.

enable

This method is called as post-receive hook in the deployment repository. It creates (or changes) the symlink new to the symlink current and restarts the application.

remote ( $command [@options] )

Run padadoy on a remote machine.

logs

Consult logfiles.

version

Show version number and exit.

DEPLOYMENT

Actually, you don't require padadoy if you only deploy at some PaaS provider, but deployment at dotCloud and OpenShift is also documented below for convenience.

On your own server

The following should work at least with a fresh Ubuntu installation and Perl >= 5.10. First you need to install git, a build toolchain, and cpanminus:

  $ sudo apt-get install git-core build-essential lbssl-dev
  $ wget -O - http://cpanmin.us | sudo perl - --self-upgrade

Now you can install padadoy from CPAN:

  $ sudo cpanm App::Padadoy

Depending on the Perl modules your application requires, you may need some additional packages, such as libexpat1-dev for XML. For instance for HTTPS you need LWP::Protocol::https that requires libnet-ssleay-perl to build:

  $ sudo apt-get install libnet-ssleay-perl
  $ sudo cpanm LWP::Protocol::https

For each deployment you create a remote repository and initialize it:

  $ padadoy init

You may then edit the file padadoy.yml to adjust the port and other settings. Back on another machine you can simply push to the deployment repository with git push. padadoy init installs some hooks in the deployment repository so new code is first tested before activation.

In most cases, you will run your application begind a reverse proxy, so you should include Plack::Middleware::XForwardedFor to get real remote IPs.

On dotCloud

Create a dotCloud account and install the command line client as documented at https://docloud.com.

On OpenShift

Create an OpenShift account, install the command line client, and create a domain, as documented at https://openshift.redhat.com/app/getting_started (you may need to sudo apt-get install libopenssl-ruby, and to find and fiddle around the client at /var/lib/gems/1.8/bin/rhc to actually make use of it). Att your OpenShift repository as remote and merge.

BACKGROUND

The remote repository contains two git hooks, which are enabled by padadoy init: the update hook calls padadoy update with the revision hash that is pushed to the repository:

    #!/bin/bash
    newrev="$3"
    padadoy update $newrev

On success, the post-receive hook calls padadoy enable to

SEE ALSO

There are many ways to deploy PSGI applications. See this presentation by Tatsuhiko Miyagawa for an overview:

http://www.slideshare.net/miyagawa/deploying-plack-web-applications-oscon-2011-8706659

By now, padadoy only supports Starman web server, but it might be easy to support more.

This should also work on Amazon EC2 but I have not tested yet. See for instance http://www.deepakg.com/prog/2011/01/deploying-a-perl-dancer-application-on-amazon-ec2/.

FAQ

What does "padadoy" mean? The funny name was derived from "PlAck DAncer DeplOYment" but it does not mean anything.

AUTHOR

Jakob Voss

COPYRIGHT AND LICENSE

This software is copyright (c) 2012 by Jakob Voss.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.