Calling reload_schedule! resets pending waiting every jobs

[soundasleep]

It seems that whenever you call Sidekiq::Scheduler.reload_schedule!, it resets all jobs scheduled for every.

This means that if you call Sidekiq::Scheduler.reload_schedule! once per hour, but you have a job that's scheduled to run every two hours (or longer), the job will never be started.

[snmgian]

Hi,

Are you modifying the schedule by calling Sidekiq.set_schedule ?

In that case calling Sidekiq::Scheduler.update_schedule instead, will work. (I'll modify the README in case you confirm)

Are you calling reload!_schedule from a sidekiq job?

[soundasleep]

I'm modifying the schedule from a sidekiq job, yes:

  def call
    Sidekiq.schedule = generated_schedule
    Sidekiq::Scheduler.reload_schedule!
    del('*', 100, false)
  end

I'm wondering if the del is getting in the way, I'll try remove that and see if that resolves the issue.

[snmgian]

I don't see del call as a problem here.

The call to reload_schedule is instantiating a new rufus scheduler, so every and interval jobs are reset.

A different method is needed to update the schedule when using Sidekiq.schedule=. I'll keep you posted.

[choubacha]

@snmgian I just ran into this issue when trying to work with the API. I think that updating the documentation would help a lot. I was able to correct it by using Sidekiq::Scheduler.instance.update_schedule after calling set_schedule when it was non-dynamic. The current documentation makes it seem like one should call reload_schedule! which ends up causing multiple firings of the same schedule.