ruby-on-rails heroku ruby-on-rails-4 rake background-process

Passed parameters to Rake not found

Following Bate's Railscast (http://railscasts.com/episodes/127-rake-in-background) on background tasks, trying it out with some mailers, not working out for me.

Application controller:

def call_rake(task, options = {})
  options[:rails_env] ||= Rails.env
  args = options.map { |n, v| "#{n.to_s}='#{v}'" }
  system "rake #{task} #{args.join(' ')} --trace 2>&1 >> #{Rails.root}/log/rake.log &"
  # need to add path to rake /usr/bin/rake, etc.
end

Request controller:

call_rake :connect_email, :requestrecord => @requestrecord, :lender_array => @lender_array, :item => item, :quantity => quantity

Rake connect_email.rb:

desc "Send email to connect borrower/lender pairs for items found"

task :connect_email => :environment do 
    RequestMailer.found_email(requestrecord, lender_array, item, quantity).deliver unless     lender_array.blank?
end

Am getting error NameError: undefined local variable or methodlender_array' for main:Object. Funny thing is, on a different mailer where the same parameters are passed except forlender_array, I get an error messageNameError: undefined local variable or method requestrecord' for main:Object.

Any thoughts on why?

And since it's my first time doing background tasks, would love to get additional insight from smart people on:

If the environment is being loaded every time this task runs, then isn't that terribly expensive? I mean that means that the environment could be re-loaded several times per hour...
Ryan mentions it's a good idea to specific the path to rake, e.g., /usr/bin/rake, how do I find what path this is on my local machine and also in production? (using Heroku)
Is it correct to say that what's happening is that Rails is queuing up a list of Rake tasks in the order that they're called by the controller and then once the request is done, just executing them one-by-one?
None of these error messages are registering in the rake.log file I created. Does it only register when the action is completed (i.e., a success)?

I'm just thinking, if person A does the action that calls the rake emails, and then person B does that same action that calls the rake emails a few miliseconds later, will Rails first execute the rake tasks for person A then do the person B request, then do person B's rake tasks, or will it do person A request, person B request, and then do the rake tasks in order received, whenever there is downtime from actual requests?

Thanks!

Solution

There are many questions in the problem, I'll first tell you the problem with your code, when you run the rake task, the code that is executed is:

task :connect_email => :environment do 
  RequestMailer.found_email(requestrecord, lender_array, item, quantity).deliver unless     lender_array.blank?
end

so when that code is executed, requesterecord, lender_array, item and quantity are not defined. That is pure ruby code. If you check the screencast in detail, what it pass to the rake task are environment variables/values, not ruby variables, the right code would be:

task :connect_email => :environment do 
  RequestMailer.found_email(ENV["requestrecord"], ENV["lender_array"], ENV["item"], ENV["quantity"]).deliver unless     lender_array.blank?
end

Now, if you do that code, it won't work either, because all the environment variables are received as String by ruby, so you have to convert them to the desired kind of variable. If you call the code:

call_rake :connect_email, lender_array: ['hello', 'world']

it will set the environment variable lender_array to '["hello", "world"]', which will be received by the connect_email task as a ruby string, if you want to convert it to a Ruby Array, you can do something like:

RequestMailer.found_email(ENV["requestrecord"], eval(ENV["lender_array"]))

Because if you eval the string '["hello", "world"]', it will be evaluated as a Ruby Array.

If the environment is being loaded every time this task runs, then isn't that terribly expensive? I mean that means that the environment could be re-loaded several times per hour...

A: Yes, every time you run a task, it will initialize rails from zero in a new process, and yes, it is expensive.

Ryan mentions it's a good idea to specific the path to rake, e.g., /usr/bin/rake, how do I find what path this is on my local machine and also in production? (using Heroku)

A: You can setup a Environment Variable for that, its pretty common to setup Environment Variables in heroku, i.e., you can setup the environment variable RAKE_PATH="/usr/bin/rake" and then on your code you can do:

system "#{ENV['RAKE_PATH'} #{task} #{args.join(' ')} --trace 2>&1 >> #{Rails.root}/log/rake.log &"

Is it correct to say that what's happening is that Rails is queuing up a list of Rake tasks in the order that they're called by the controller and then once the request is done, just executing them one-by-one?

A: No, every time you run a rake task, is a different process being created to run that specific task, not queue at all.

None of these error messages are registering in the rake.log file I created. Does it only register when the action is completed (i.e., a success)?

A: Rake tasks usually logs information, I'm not sure why that exception is not being logged dough, but I usually check the logs and I see the rake tasks running and logging to them.

In order to be able to perform better your tasks on background, you can check delayed_job, which stores the jobs to be run on a database, and then a daemon fetch the tasks and run them in background.