Errors are not stored into last_error_backtrace/last_error_message if job maximum_retry_count == 0

[rolftimmermans]

I have the following job. It must only be attempted once.

class RunPortalAgent < Que::Job
  self.maximum_retry_count = 0

  def run(session_id)
    session = Portal::Session.with_pk(session_id)
    if session.present?
      # Code that can raise error on failure
    end

    destroy
  end
end

When this code fails, the following row is added in the database. Note the absence of last_error_message/last_error_backtrace.

{
  "priority": 100,
  "run_at": "2025-10-13 14:04:25.785659+02",
  "id": 1,
  "job_class": "RunPortalAgent",
  "error_count": 1,
  "last_error_message": null,
  "queue": "default",
  "last_error_backtrace": null,
  "finished_at": null,
  "expired_at": "2025-10-13 14:04:25.899913+02",
  "args": [
    "5d4296f2-2891-445f-ad95-6d379b332e11"
  ],
  "data": {
  },
  "job_schema_version": 2,
  "kwargs": {
  }
}

It seems that que incorrectly only saves the error if it is going to retry the job?
See:

que/lib/que/worker.rb

Lines 155 to 171 in 17fb2c3

	if max_retry_count && error_count > max_retry_count
	Que.execute :expire_job, [job.fetch(:id)]
	else
	delay =
	job_class.
	resolve_que_setting(
	:retry_interval,
	error_count,
	)
	Que.execute :set_error, [
	delay,
	"#{error.class}: #{error.message}".slice(0, 500),
	(error.backtrace || []).join("\n").slice(0, 10000),
	job.fetch(:id),
	]
	end

To me it would seem logical to always save the error on failure, regardless whether the job will be expired or retried.

[ZimbiX]

Wow, thanks for the report. I'm surprised I never noticed that. I agree with your assessment.

I know I haven't been around much for a while, but I'd be happy to accept a PR for this.