Log from within collectiveidea/delayed_job

Question

I try to log from within a delayed_job in rails.

I configure it as following:

Delayed::Worker.destroy_failed_jobs = false
Delayed::Worker.max_attempts = 3
Delayed::Worker.backend = :active_record
Delayed::Worker.logger = ActiveSupport::BufferedLogger.new("log/#{Rails.env}_delayed_jobs.log", Rails.logger.level)
Delayed::Worker.logger.auto_flushing = 1

Define my job:

class TestJob

    def initialize(user)
        @user = user
    end


     #called when enqueue is performed
    def enqueue(job)
       Delayed::Worker.logger.info("TestJob: enqueue was called")
    end

    def perform
        Delayed::Worker.logger.info("\n\n\n\nTestJob: in perform, for user #{@user.twitter_username}")
    end
end

But when I call enquee on my job

 Delayed::Job.enqueue(TestJob.new(user), 2)

The log files remain empty, even though the delayed_jobs table shows that the job was performed.

Any ideas?

It appears you're not alone: http://stackoverflow.com/questions/3500200/getting-delayed-job-to-log — Dave Rapin, May 02 '11 at 18:09

score 0 · Answer 1 · answered May 12 '11 at 06:06

Try it without using DJ's namespace.

Here's code from a job I was using a while ago. Note: Logger.new will just append on to a .log file if it already exists.

class ScrapingJob < Struct.new(:keyword_id)
  def perform
      begin
       ...
      rescue =>  error
        log = Logger.new("#{Rails.root}/log/scraping_errors.log")
        log.debug "logger created for '#{keyword.text}' (#{keyword.id}) on '#{keyword.website.url}' (#{keyword.website.id})"
        log.debug "time: #{measured_at.to_s(:short)}"
        log.debug "error: #{error}"
        log.debug ""
      end
    end
  end

score 0 · Answer 2 · answered Jun 14 '11 at 16:09

My guess it is a serialization issue with DJ trying to serialize the User object. DJ does a bad job of raising those errors.

Are you using Mongoid by any chance?? I know there are issues there with JSONizing a Mongoid object.

Does it work if you pass the user_id string to the Job, instead of the User object? like so ...

class TestJob

    def initialize(user_id)
        @user = User.find(user_id)
    end

    # ...

end

Even though it is not as pretty, I think it is best practice to pass literals (Strings, Floats, etc...) to DelayedJob or any background worker. This makes it easier to swap processers in case you want to migrate to Resque in the future.

Log from within collectiveidea/delayed_job

2 Answers2