Will sidekiq_retries_exhausted be called before a job disappears if dead: false or will it only be called before moving a job to the deadset?

[daugaard]

I am asking the question because the wiki (https://github.com/sidekiq/sidekiq/wiki/Error-Handling#configuration) says:

After retrying so many times, Sidekiq will call the sidekiq_retries_exhausted hook on your Job if you've defined it. The hook receives the queued job hash as an argument and is called right before Sidekiq moves the job to the Dead set.

But from reading the code

sidekiq/lib/sidekiq/job_retry.rb

Lines 240 to 276 in e798c23

	def retries_exhausted(jobinst, msg, exception)
	rv = begin
	block = jobinst&.sidekiq_retries_exhausted_block
	# the sidekiq_retries_exhausted_block can be defined in a wrapped class (ActiveJob for instance)
	unless msg["wrapped"].nil?
	wrapped = Object.const_get(msg["wrapped"])
	block = wrapped.respond_to?(:sidekiq_retries_exhausted_block) ? wrapped.sidekiq_retries_exhausted_block : nil
	end
	block&.call(msg, exception)
	rescue => e
	handle_exception(e, {context: "Error calling retries_exhausted", job: msg})
	end
	return if rv == :discard # poof!
	send_to_morgue(msg) unless msg["dead"] == false
	@capsule.config.death_handlers.each do |handler|
	handler.call(msg, exception)
	rescue => e
	handle_exception(e, {context: "Error calling death handler", job: msg})
	end
	end
	def send_to_morgue(msg)
	logger.info { "Adding dead #{msg["class"]} job #{msg["jid"]}" }
	payload = Sidekiq.dump_json(msg)
	now = Time.now.to_f
	redis do |conn|
	conn.multi do |xa|
	xa.zadd("dead", now.to_s, payload)
	xa.zremrangebyscore("dead", "-inf", now - @capsule.config[:dead_timeout_in_seconds])
	xa.zremrangebyrank("dead", 0, - @capsule.config[:dead_max_jobs])
	end
	end
	end

I get the impression that it is already called regardless and the decision whether to send it to the dead set or not happens after this has been called.

[mperham]

Your impression (and the code) are correct. The wiki page is awkwardly phrased.