Best Practice For Background Processes

[John Kopanas]

I have a couple background processes that run from anywhere between 2 seconds and 30 seconds that run periodically (some run every 15 minutes, some every 4 hrs).  To run them I call a webpage using curl in a cron job.  I have a funny feeling this is not the best thing to do :-).

All the tasks I have manipulate the DB.  Either re-weighing users or copying data from one table to another.

What are the best practices to run these types of periodic tasks? 

Thanks before hand for your feedback!

Your friend,

John

--
John Kopanas
jo...@kopanas.com

Blog: http://www.kopanas.com
Conference: http://www.cusec.net
Twits: http://www.twitter.com/kopanas

[Tim McIntyre]

Here is what I do:

I have a file lib/offline/model_loader.rb that looks like this:

module ModelLoader

def load_model(model_path=nil, database_config=nil, environment=nil)
model_path ||= "/../../app/models/"
database_config ||= "/../../config/database.yml"
environment ||= "development"
$LOAD_PATH << File.expand_path(File.dirname(__FILE__) +
model_path)
require "yaml"
require "rubygems"
gem "activerecord"
require File.dirname(__FILE__) + '/../../config/environment'
conf = YAML::load(File.open(File.dirname(__FILE__) +
database_config))
ActiveRecord::Base.establish_connection(conf[environment])
end

end

Then in whatever script you wanna run say script/my_site_name/
some_script_for_cron.rb

#!/usr/bin/env ruby

rootd = File.dirname(__FILE__) + '/../../'
require rootd + 'lib/offline/model_loader'
include ModelLoader
ModelLoader::load_model(nil, nil, 'development')

# now you have access to all your models

MyModel.find(:all).each do |instance|
...
...
end

There are other situations that may call for access to controllers/
views, there's a solution for that as well, but I suspect this is what
you need.

Hope that helps, good luck!
Tim

On Apr 30, 1:03 pm, "John Kopanas" <kopa...@gmail.com> wrote:
> I have a couple background processes that run from anywhere between 2
> seconds and 30 seconds that run periodically (some run every 15 minutes,
> some every 4 hrs). To run them I call a webpage using curl in a cron job.
> I have a funny feeling this is not the best thing to do :-).
> All the tasks I have manipulate the DB. Either re-weighing users or copying
> data from one table to another.
>
> What are the best practices to run these types of periodic tasks?
>
> Thanks before hand for your feedback!
>
> Your friend,
>
> John
>
> --
> John Kopanas

> j...@kopanas.com

[Andrew Bloom]

this is a perfect place for rake!

i encounter situations like this all the time at work. simply take the
code from your action that you are hitting with curl and wrap them in
a rake task. then set up your cron job to do something along the lines
of:

15 2 * * * /usr/local/bin/rake -f /var/www/site/Rakefile do:something
RAILS_ENV=production

and in lib/tasks/something.rake:

namespace :do do
desc "does something"
task :something => :environment do
<code moved from the action in your controller to here>
end
end