why use wait_for_termination method will stuck the code

Start with the thread_pool.shutdown and thread_pool.wait_for_termination calls in the issue's example, then investigate the worker's ClassFactory.create_dynamic_model call under Ruby with concurrent-ruby 1.2.0 and no concurrent-ruby-ext. Reproduce the hang on Linux and establish why the worker does not finish; done means the supplied workload terminates without leaving wait_for_termination stuck.

        user_ids,org,delete_restore = get_org_and_user(delete_restore_id)
        return unless user_ids.present?
        work_num = [user_ids.length,50].min
        thread_pool = Concurrent::FixedThreadPool.new(work_num)
        error_map = Concurrent::Map.new
        redis_key = RedisClient::DDRS_BULK_DELETE_KEY % org.id  ## TODO
        set_redis_total_num(redis_key,user_ids.length)
  
        user_ids.each do |id|
          thread_pool.post do
            #delete_single_user(id,error_map,redis_key)
            target_class = ClassFactory.create_dynamic_model("zendmodo", "privacy_settings", true)
            puts id
            puts target_class
          end
        end

       thread_pool.shutdown
       thread_pool.wait_for_termination

the wait_for_termination method will stuck the following code

target_class = ClassFactory.create_dynamic_model("zendmodo", "privacy_settings", true)

• Operating system: linux
• Ruby implementation: Ruby
• concurrent-ruby version: 1.2.0
• concurrent-ruby-ext installed: no
• concurrent-ruby-edge used: no