Skip to content

fhwang/auto_cron

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

= AutoCron

AutoCron is a Rails plugin that lets you write your crontab as files checked in with your Rails project, and writes your crontab every time you deploy. This way, the crontab history is tracked alongside the rest of your application, and backed up in your source control.

Please note a few important caveats:

* This is an all-or-nothing solution. If you use AutoCron, you can't do any editing of crontabs by hand, because AutoCron will overwrite those changes on your next deploy.

* Similarly, if you have more than one Rails app on a server that uses AutoCron, they can't have the same user account on the server, or else they'll overwrite each others' crontabs. One solution worth trying would be to give each app a different user account, with a common group.

== Installation

EDGE Rails has support for plugin installation from Git, so if you're using EDGE Rails you can install with:

  ./script/plugin install git://github.com/fhwang/auto_cron.git

We're maintaining an SVN clone of the plugin for those of you who aren't on EDGE Rails yet:

  ./script/plugin install svn://rubyforge.org/var/svn/nycrb/auto_cron

== Integrating AutoCron with your application

There are two steps to integrating AutoCron with your application:

  1. Write your auto_cron templates

  2. Call the auto_cron tasks from inside your deployment script

== Modifying your auto_cron templates

By default, AutoCron creates a folder config/auto_cron for you, and puts three templates there.

You'll most likely want to edit publish.erb: That will run on one machine and is a good place for most of your cronjobs.

If you have any jobs that need to run on every machine in your infrastructure, you'll want to put those into app.erb. (See "The 'publish' role vs. the 'app' role" for more on this.)

There's also a header.erb file, which is just a warning that's published at the top of every crontab on every server. You probably don't need to edit this.

Since these are .erb files, you can do lots of things with them:

  # Update friends count once an hour
  0 * * * * cd /usr/local/apps/myrailsapp/current; /usr/local/bin/ruby ./script/runner -e <%= RAILS_ENV %> "User.update_friends_count"

  # Send emails every minute, but only in production
  <% if RAILS_ENV == 'production' %>
  * * * * * cd /usr/local/apps/myrailsapp/current; RAILS_ENV=<%= RAILS_ENV %> ./vendor/ar_mailer/bin/ar_sendmail
  <% end %>
  
== The "publish" role vs. the "app" role

Whether you're running on one server or on 100, you probably have a handful of jobs that only need to be run by one machine. For example, you might do an expensive precalculation once a day, and save the results into the database where all the app servers can get them as needed. AutoCron calls that server the "publish" server, since its job is mostly to publish calculated data into caches. Of course, that's not all the "publish" server does, it can also send emails, or fire off system maintenance tasks. The important distinction is that there should only be one "publish" server in your entire infrastructure.

If you have any jobs that need to run on every machine in your infrastructure, you'll want to put them into app.erb. For example, maybe your Rails app will touch local files in a way that needs to be cleaned up periodically. However, this is a lot less common.

== Integrating with Capistrano 2

To integrate auto_cron with Capistrano, first create a :publish role for each of your environments. This should only be one server per environment.

  if rails_env == 'staging'
    role :app,     'staging.myrailsapp.com'
    role :publish, 'staging.myrailsapp.com'
  elsif rails_env == 'production'
    role :app,     'prod01.myrailsapp.com', 'prod02.myrailsapp.com',
                   'prod03.myrailsapp.com'
    role :publish, 'prod01.myrailsapp.com'
  end

To update your live crontabs after your deploy, enter these two lines in your Capistrano script.

  auto_cron_app
  auto_cron_publish

They should run sometime after your new code is deployed; we put this in the :after_symlink task.

auto_cron_app should always be called before auto_cron_publish.

OR

  if rails_env == 'staging'
    role :publish_cron, 'staging.myrailsapp.com', :crontab => 'publish'
  
  elsif rails_env == 'production'
    role :publish_cron, 'prod01.myrailsapp.com', 'prod02.myrailsapp.com', :crontab => "app"
    role :publish_cron, 'cron.myrailsapp.com', :crontab => 'publish'
  
  end

Then use this task somewhere in your deployment script

  publish_crons

e.g.

task :after_symlink do
  transaction do
    publish_crons
  end
end

== Integrating with Vlad the Deployer

Since we don't yet use Vlad at Diversion, we haven't had a chance to try to use auto_cron with it, so you're on your own here. Sorry! If you get it working, though, please let us know and we'll add your instructions.

== Contact

If you have any bugs, questions, etc., please feel to email me:

Francis Hwang
francis@diversionmedia.com



Copyright (c) 2008 Diversion Media, released under the MIT license

About

Rails plugin that lets you save your crontab in your project, which is then published to your servers every time you deploy.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages