Maybe there is deadlock in ~rcp_server()...

[objects76]

I think the lock(std::uniquelock lock(mtx)) makes deadlock in checkthread...

        ~rpc_server() {
            {
                //std::unique_lock<std::mutex> lock(mtx_);//-jjkim
                stop_check_ = true;
                cv_.notify_all();
                check_thread_->join();
            }

            {
                std::unique_lock<std::mutex> lock(sub_mtx_);
                stop_check_pub_sub_ = true;
                                     lock.unlock();//+jjkim
                sub_cv_.notify_all();
                                    pub_sub_thread_->join();
            }

            io_service_pool_.stop();
            if(thd_){
                thd_->join();
            }
        }

[qicosmos]

I think it's right, no deadlock here.