Disable automatic retry with ActiveJob, used with Sidekiq

Question

Is there a way to disable automatic retry with ActiveJob and Sidekiq ?

I know that with Sidekiq only, we just have to put

sidekiq_options :retry => f         


        

Answer 1

As of sidekiq 6.0.1, it is possible to pass the following to an ActiveJob worker to prevent it from retrying:

class ExampleJob < ActiveJob::Base
  sidekiq_options retry: false

  def perform(*args)
    # Perform Job
  end
end

More information: https://github.com/mperham/sidekiq/wiki/Active-Job#customizing-error-handling

EDIT:

According to this this requires Rails 6.0.1 or later as well.

Answer 2

I had this same need, ie ActiveJob wrapping Sidekiq but wanting to support max_retries. I put this in an initializer. If #max_retries is defined on an ActiveJob job, it will be used to set retries. If #ephemeral? is defined and returns true, job will not be rerun and will not be transferred to 'dead' if it fails.

class Foobar::SidekiqClientMiddleware
  def call(worker_class, msg, queue, redis_pool)
    aj_job = ActiveJob::Base.deserialize(msg['args'][0]) rescue nil
    msg['retry'] = aj_job.respond_to?(:max_retries) ? aj_job.max_retries : 5
    msg['retry'] = false if aj_job.respond_to?(:ephemeral?) && aj_job.ephemeral?
    yield
  end
end

Sidekiq.configure_client do |config|
  config.redis = { url: "redis://#{redis_host}:6379/12" }
  config.client_middleware do |chain|
    chain.add Foobar::SidekiqClientMiddleware
  end
end

Sidekiq.configure_server do |config|
  config.redis = { url: "redis://#{redis_host}:6379/12" }
  config.client_middleware do |chain|
    chain.add Foobar::SidekiqClientMiddleware
  end
end

Note: it actually is important to add this to the middleware chain for both client and server if any of your jobs create new jobs themselves as they are executed.

Answer 3

If you want to disable retry (or add any other Sidekiq options) for ActiveJob from gems (like for ActionMailbox::RoutingJob), you can use this approach (Rails 6.0.2+).

1) Create a module with the desired Sidekiq options (using ActiveSupport::Concern)

# lib/fixes/action_mailbox_routing_job_sidekiq_fix.rb

module ActionMailboxRoutingJobSidekiqFix
  extend ActiveSupport::Concern

  included do
    sidekiq_options retry: false
  end
end

2) Include it in the job class in an initializer.

# config/initializers/extensions.rb

require Rails.root.join('lib', 'fixes', 'action_mailbox_routing_job_sidekiq_fix')

Rails.configuration.to_prepare do
  ActionMailbox::RoutingJob.include ::ActionMailboxRoutingJobSidekiqFix
end

Answer 4

You can catch up the exception and to do nothing instead retry or to configure retry:

  class ExampleJob < ActiveJob::Base
    rescue_from(StandardError) do |exception|
      Rails.logger.error "[#{self.class.name}] Hey, something was wrong with you job #{exception.to_s}"       
    end

    def perform
      raise StandardError, "error_message"
    end
  end

  class ExampleJob < ActiveJob::Base
    rescue_from(StandardError) do |exception|
      retry_job wait: 5.minutes, queue: :low_priority     
    end

    def perform
      raise StandardError, "error_message"
    end
  end

For running retrying you can use retry_on method retry_on method doc

Sidekiq wiki for retries with Active Job integration

Answer 5

There is no way to configure anything about Sidekiq with ActiveJob. Use a Sidekiq Worker if you don't want to use the defaults.

Answer 6

Ok thanks for the answer.

Just for information, I also asked the question in an issue related to this subject on ActiveJob Github repository : https://github.com/rails/activejob/issues/47

DHH answered me a solution I haven't tested but that can do the job.

Personnally, I finally put this in an initializer in order to disable Sidekiq retries globally and it works well :

Sidekiq.configure_server do |config|
   config.server_middleware do |chain|
      chain.add Sidekiq::Middleware::Server::RetryJobs, :max_retries => 0
   end
end