weird disconnection and reconnection

[faithware]

My two node cluster is initialized this way

  ptr<state_machine> my_state_machine = cs_new<echo_state_machine>();
  ptr<state_mgr> my_state_manager =
      cs_new<inmem_state_mgr>(cluster_id, def_endpoint);

  asio_service::options asio_opt;
  asio_opt.thread_pool_size_ = 4;

  // Raft parameters.
  raft_params params;
  // heartbeat: 100 ms, election timeout: 200 - 400 ms.
  params.heart_beat_interval_ = 100;
  params.election_timeout_lower_bound_ = 200;
  params.election_timeout_upper_bound_ = 400;
  //params.auto_adjust_quorum_for_small_cluster_ = true;
  // Upto 5 logs will be preserved ahead the last snapshot.
  params.reserved_log_items_ = 5;
  // Snapshot will be created for every 5 log appends.
  params.snapshot_distance_ = 5;
  // Client timeout: 3000 ms.
  params.client_req_timeout_ = 3000;
  // According to this method, `append_log` function
  // should be handled differently.
  params.return_method_ = raft_params::blocking;

  // Initialize Raft server listening on port 12345.
  // It will organize a single-node Raft cluster.
  raft_launcher launcher;

  server =
      launcher.init(my_state_machine, my_state_manager, my_logger,
                    net_config.clustering_port, asio_opt, params, user_init);

When I launch I set cluster ID3 to be the leader and it adds the cluster ID1 , after that, the cluster renegociate the leader and cluster ID 1 becomes leader.
After that, cluster ID 1 (the new leader) is stuck all the time in

 Connection opened 
I am a leader
new session from leader127.0.0.1
Connection closed leader:1peer :-1
new session from leader192.168.3.107
 Connection opened 
new session from leader127.0.0.1
Connection closed leader:1peer :-1
 Connection opened 
new session from leader127.0.0.1
Connection closed leader:1peer :-1
 Connection opened 

And it remains like this for ever .
Is this normal? I thought ones everything is set , the leader doesn't disconnect and reconnects to itself?

[greensky00]

Hi @faithware
This doesn't seem normal. Can you share the log file of both servers?

[faithware]

It turned out to be a network issue with my wifi card. Is there any clear way to restart the raft server?

[greensky00]

@faithware
If you meant restarting it without killing the process, please refer to this test code to shut it down and then restart it:

NuRaft/tests/unit/asio_service_test.cxx

Lines 2297 to 2298 in 789cc75

	s3.raftServer->shutdown();
	s3.stopAsio();

Call raft_server::shutdown(), and then close Asio like this:

NuRaft/tests/unit/raft_package_asio.hxx

Lines 188 to 200 in 789cc75

	if (asioListener) {
	asioListener->stop();
	asioListener->shutdown();
	}
	if (asioSvc) {
	asioSvc->stop();
	size_t count = 0;
	while (asioSvc->get_active_workers() && count < 500) {
	// 10ms per tick.
	timer_helper::sleep_ms(10);
	count++;
	}
	}

[faithware]

thanks @greensky00
I just used server->shutdown and restarted the raft initialization again and it works.
Do I have to explicitly kill asio even though my intent is to restart the server?

[greensky00]

@faithware You don't have to, but just to be safe. :)