Making rufus-scheduler jobs persist after a server reboot

The Application

I am writing an application that runs the scheduler and fetches the records after someone saves an entry into the database.

initializers/scheduler.rb

require 'rufus-scheduler' 


SCHEDULER = Rufus::Scheduler.new(:lockfile => ".rufus-scheduler.lock")
unless SCHEDULER.down?

  # SCHEDULER.every '3m', :tags => 121 do
  #   Rails.logger.info "Ping!!"
  #     Rails.logger.flush
  # end

end

controllers/abc_controller.rb

class AbcController < ApplicationController

  def new
    obj = ModelClass.new(params)
    if obj.save
      SCHEDULER.every obj.frequency do
        # run a job
      end
    end
  end

end

Problem

The job is started from the controller, but the problem is when a server is restarted the job is lost.

I have read the rufus-scheduler's readme file and it is mentioned that it is a non-feature. But still I want to know if there is any solution to this (persisting the jobs after a server reboot) or do I have to change my approach?

P.S. I searched for it on stack overflow but I am sorry if I missed anything or missed a similar question.

EDIT

For all those who want to run the scheduler from the controller itself using params, think of a strategy to save the data in the db(temp table or something) if possible and then bind the create event with scheduling. Re-schedule the jobs on scheduler initialization, i.e, server reboot as mentioned in the answer by @jmettraux. Hope this helps.

Solution

You are right, rufus-scheduler is not tied to any persistence.

Since you're already tying records to schedules (by scheduling a job for each object getting created), why don't you simply reschedule when your application restarts?

Something like (pseudo-code aligned on yours):

require 'rufus-scheduler'


SCHEDULER = Rufus::Scheduler.new(:lockfile => ".rufus-scheduler.lock")

unless SCHEDULER.down?

  ModelClass.all.each do |obj|
    SCHEDULER.every obj.frequency do
      # the job
    end
  end
end