Error Callbacks and Notifications

Introduction

When jobs fail permanently, you need to know. Active Job provides rescue_from and after_discard callbacks for custom error handling.

Key Concepts

• rescue_from: Catches exceptions for custom handling. Re-raise to allow retries.
• after_discard: Runs after a job is permanently discarded.

Real World Context

A payment job that exhausts retries needs immediate attention. An import that fails on one record should log and continue.

Deep Dive

rescue_from

ruby
class CriticalJob < ApplicationJob
  retry_on StandardError, attempts: 3

  rescue_from StandardError do |exception|
    Rails.logger.error "Job failed: #{exception.message}"
    ErrorTracker.notify(exception)
    raise exception  # Re-raise to trigger retry_on
  end

  def perform(data)
    # Job logic
  end
end

after_discard

ruby
class PaymentJob < ApplicationJob
  retry_on PaymentGateway::Error, attempts: 5

  after_discard do |job, exception|
    AdminMailer.job_permanently_failed(
      job_class: job.class.name,
      arguments: job.arguments,
      error: exception&.message
    ).deliver_later(queue: :critical)
  end

  def perform(payment_id)
    Payment.find(payment_id).then { |p| PaymentGateway.charge(p) }
  end
end

Common Pitfalls

1. Forgetting to re-raise in rescue_from — Without re-raising, the job is considered successful.
2. Sending failure notifications on the same backed-up queue — Use a separate :critical queue.

Best Practices

1. Integrate error tracking — Sentry, Honeybadger, etc.
2. Update affected records on failure — Set a failed status so dashboards show what happened.

Summary

• rescue_from handles exceptions — re-raise to allow retries.
• after_discard runs when all retries are exhausted.
• Integrate error tracking for automated alerting.
• Update records to reflect failure status.

class ImportJob < ApplicationJob
  retry_on StandardError, attempts: 3

  after_discard do |job, exception|
    import = Import.find(job.arguments.first)
    import.update!(status: :failed, error: exception&.message)
    AdminNotifier.alert("Import #{import.id} permanently failed")
  end

  def perform(import_id)
    Import.find(import_id).then { |i| ImportService.process(i) }
  end
end