Manually Retry Job in Delayed_job

Question

Delayed::Job's auto-retry feature is great, but there's a job that I want to manually retry now. Is there a method I can call on the job itself like...

Delayed::Job.all[0].perform

or run, or something. I tried a few things, and combed the documentation, but couldn't figure out how to execute a manual retry of a job.

`Delayed::Worker.new.run(Delayed::Job.first)` [ref](http://stackoverflow.com/a/20146200/495132) — Jake Berger, Jan 06 '15 at 11:12

The Who · Accepted Answer · 2012-06-14T15:35:17.340

115

To manually call a job

Delayed::Job.find(10).invoke_job # 10 is the job.id

This does not remove the job if it is run successfully. You need to remove it manually:

Delayed::Job.find(10).destroy

edited Jun 14 '12 at 15:35

answered Oct 26 '10 at 03:30

The Who

6,552
5
36
33

10

The alternative suggested by @joe is a safer better, especially if the job needs to know whether it is running in script/console or inside a job runner. Try this to queue the job for immediate retry Delayed::Job.first.update_attributes(:attempts=>0, :run_at=>Time.now, :failed_at => nil, :locked_by=>nil, :locked_at=>nil) – so_mv Mar 10 '12 at 00:16
8

`attempts` cannot be assigned in `update_attributes` because it is a protected attribute. I just do: `dj = Delayed::Job.first; dj.run_at = Time.now; dj.attempts = 0; dj.save!;` – Anjan Aug 20 '12 at 11:00
To do it en-masse this worked (approx 100 jobs) Delayed::Job.where.all.each {|dj| dj.run_at = Time.now; dj.attempts = 0; dj.save!} – tobinharris Mar 24 '14 at 13:22
1

I just do `Delayed::Job.where.not(last_error: nil).update_all(run_at: Time.current)` – Jason Galuten May 16 '18 at 20:34
this doesn't set the `last_error` attribute if the job fails – localhostdotdev Apr 18 '19 at 19:01

score 40 · Answer 2 · edited Sep 09 '15 at 11:59

40

Delayed::Worker.new.run(Delayed::Job.last)

This will remove the job after it is done.

edited Sep 09 '15 at 11:59

Sergio Tulentsev

226,338
43
373
367

answered Sep 04 '15 at 08:41

Beena Shetty

3,676
2
28
31

It will remove it even if it fails – aledustet Dec 08 '16 at 21:19
1

For all the delayed job you can do `Delayed::Job.find_each(batch_size: 100) { |d| Delayed::Worker.new.run(d) }` – MatayoshiMariano Mar 10 '17 at 13:41
1

@aledustet Surely not. Are you positive? – Joshua Pinter Jan 25 '19 at 00:53
This worked for me, for a job that did not fail. Didn't try one that fails, but I really doubt it will remove it, that would be extremely surprising. – dzajic Jan 23 '20 at 10:47

score 12 · Answer 3 · answered Jul 22 '10 at 21:16

12

You can do it exactly the way you said, by finding the job and running perform.

However, what I generally do is just set the run_at back so the job processor picks it up again.

answered Jul 22 '10 at 21:16

Joe Martinez

854
4
9

1

There is no `perform` method for delayed job object. The closest is `Delayed::Job.find(10).payload_object.perform`, and one should not use that. – lulalala Jun 14 '12 at 10:05

score 10 · Answer 4 · answered Sep 07 '12 at 01:52

10

I have a method in a controller for testing purposes that just resets all delayed jobs when I hit a URL. Not super elegant but works great for me:

# For testing purposes
  def reset_all_jobs
    Delayed::Job.all.each do |dj|
      dj.run_at = Time.now - 1.day
      dj.locked_at = nil
      dj.locked_by = nil
      dj.attempts = 0
      dj.last_error = nil
      dj.save
    end
    head :ok
  end

answered Sep 07 '12 at 01:52

Tony

18,776
31
129
193

1

You can't use update_all ebcause attempts are protected attribute – Ivailo Bardarov Mar 13 '13 at 13:31
3

update_all uses SQL directly and thus does not invoke validations (or protected attributes etc.) – Michael May 04 '16 at 13:06

score 8 · Answer 5 · edited Oct 18 '16 at 12:01

8

Prior answers above might be out of date. I found I needed to set failed_at, locked_by, and locked_at to nil:

(for each job you want to retry):

d.last_error = nil
d.run_at = Time.now
d.failed_at = nil
d.locked_at = nil
d.locked_by = nil
d.attempts = 0
d.failed_at = nil # needed in Rails 5 / delayed_job (4.1.2)
d.save!

edited Oct 18 '16 at 12:01

David Silva Smith

11,498
11
67
91

answered Mar 21 '13 at 15:30

jpw

18,697
25
111
187

score 5 · Answer 6 · answered Jan 30 '16 at 12:14

if you have failed delayed job which you need to re-run, then you will need to only select them and set everything refer to failed retry to null:

Delayed::Job.where("last_error is not null").each do |dj|
  dj.run_at = Time.now.advance(seconds: 5)
  dj.locked_at = nil
  dj.locked_by = nil
  dj.attempts = 0
  dj.last_error = nil
  dj.failed_at = nil
  dj.save  
end

score 4 · Answer 7 · answered Feb 28 '13 at 15:16

4

In a development environment, through rails console, following Joe Martinez's suggestion, a good way to retry all your delayed jobs is:

Delayed::Job.all.each{|d| d.run_at = Time.now; d.save!}

answered Feb 28 '13 at 15:16

Darme

6,984
5
37
52

6

Updating `run_at` in 4.0.1 does not appear to be enough. I had to do the following: `Delayed::Job.where("failed_at is not null").each do |dj| dj.run_at = Time.now; dj.last_error = nil; dj.failed_at = nil; dj.save! end` – steakchaser May 13 '14 at 18:43

score 3 · Answer 8 · answered Jun 09 '15 at 08:53

3

Delayed::Job.all.each(&:invoke_job)

answered Jun 09 '15 at 08:53

rubycop

31
1

score 2 · Answer 9 · answered Dec 29 '17 at 18:08

2

Put this in an initializer file!

module Delayed
  module Backend
    module ActiveRecord
      class Job
        def retry!
          self.run_at = Time.now - 1.day
          self.locked_at = nil
          self.locked_by = nil
          self.attempts = 0
          self.last_error = nil
          self.failed_at = nil
          self.save!
        end
      end
    end
  end
end

Then you can run Delayed::Job.find(1234).retry!

This will basically stick the job back into the queue and process it normally.

answered Dec 29 '17 at 18:08

Blair Anderson

19,463
8
77
114

1

Yeah, this is a clean way of putting it back in the queue to be processed and letting the workers process it normally. I would only suggest not having `Time.now - 1.day` and just using `Time.now` or `Time.zone.now`. No sense in putting it back a day (except maybe for priority) and it can add confusion if looking at this later on. – Joshua Pinter Dec 07 '21 at 17:46

Manually Retry Job in Delayed_job

9 Answers9

Linked