[SOLVED] GatewayESP8266MQTTClient - Recovery failure after gateway outage

[cimba007]

Hello,

I am using the "GatewayESP8266MQTTClient" and "LightSensor" Sketch with default settings.

My System: MySensorsNet 2.0.0, Arduino IDE 1.6.10 hourly (22.07.2016), Windows 8.1 64bit

I tried to find out how robust this setup is so i forcefully reset the Gateway (power off) and waited until the "Light Sensor" node entered failure state:

**
"Light Sensor"-Node**

!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
!TSP:SEND:TNR<\n>
vcc: 3280<\r><\n>
511, 1017<\r><\n>
!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
!TSP:SEND:TNR<\n>
vcc: 3280<\r><\n>
511, 1015<\r><\n>
!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
!TSP:SEND:TNR<\n>
vcc: 3280<\r><\n>
511, 1017<\r><\n>
!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
!TSP:SEND:TNR<\n>
vcc: 3280<\r><\n>
511, 1015<\r><\n>
!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
TSM:FPAR<\n>
TSP:MSG:SEND 10-10-255-255 s=255,c=3,t=7,pt=0,l=0,sg=0,ft=0,st=bc:<\n>
!TSP:SEND:TNR<\n>
vcc: 3280<\r><\n>
511, 1018<\r><\n>
!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
!TSP:SEND:TNR<\n>
vcc: 3280<\r><\n>
511, 1017<\r><\n>
!TSP:SEND:TNR<\n>
Send failed!<\r><\n>
!TSP:SEND:TNR

After the Gateway is up again the "Light Sensor"-Node still can't find back the Gateway

And the gateways is somehow doing nothing .. at least I do not understand what it is doing.

.scandone<\n>
state: 0 -> 2 (b0)<\n>
state: 2 -> 3 (0)<\n>
state: 3 -> 5 (10)<\n>
add 0<\n>
aid 1<\n>
cnt <\n>
<\n>
connected with FRITZ!Box Fon WLAN 7390, channel 11<\n>
dhcp client start...<\n>
.............ip:192.168.178.67,mask:255.255.255.0,gw:192.168.178.1<\n>
.IP: 192.168.178.67<\r><\n>
0;255;3;0;9;No registration required<\n>
0;255;3;0;9;Init complete, id=0, parent=0, distance=0, registration=1<\n>
IP: 192.168.178.67<\r><\n>
0;255;3;0;9;Attempting MQTT connection...<\n>
0;255;3;0;9;MQTT connected<\n>
0;255;3;0;9;TSP:MSG:READ 10-10-255 s=255,c=3,t=7,pt=0,l=0,sg=0:<\n>
0;255;3;0;9;TSP:MSG:BC<\n>
0;255;3;0;9;TSP:MSG:FPAR REQ (sender=10)<\n>
0;255;3;0;9;TSP:CHKUPL:OK (FLDCTRL)<\n>
0;255;3;0;9;TSP:MSG:GWL OK<\n>
0;255;3;0;9;!TSP:MSG:SEND 0-0-10-10 s=255,c=3,t=8,pt=1,l=1,sg=0,ft=0,st=fail:0<\n>
pm open,type:2 0<\n>
0;255;3;0;9;TSP:MSG:READ 10-10-255 s=255,c=3,t=7,pt=0,l=0,sg=0:<\n>
0;255;3;0;9;TSP:MSG:BC<\n>
0;255;3;0;9;TSP:MSG:FPAR REQ (sender=10)<\n>
0;255;3;0;9;TSP:CHKUPL:OK<\n>
0;255;3;0;9;TSP:MSG:GWL OK<\n>
0;255;3;0;9;!TSP:MSG:SEND 0-0-10-10 s=255,c=3,t=8,pt=1,l=1,sg=0,ft=0,st=fail:0<\n>
0;255;3;0;9;TSP:MSG:READ 10-10-255 s=255,c=3,t=7,pt=0,l=0,sg=0:<\n>
0;255;3;0;9;TSP:MSG:BC<\n>
0;255;3;0;9;TSP:MSG:FPAR REQ (sender=10)<\n>
0;255;3;0;9;TSP:CHKUPL:OK<\n>
0;255;3;0;9;TSP:MSG:GWL OK<\n>
0;255;3;0;9;!TSP:MSG:SEND 0-0-10-10 s=255,c=3,t=8,pt=1,l=1,sg=0,ft=0,st=fail:0<\n>
0;255;3;0;9;TSP:SANCHK:OK<\n>
0;255;3;0;9;TSP:SANCHK:OK<\n>
0;255;3;0;9;TSP:MSG:READ 10-10-255 s=255,c=3,t=7,pt=0,l=0,sg=0:<\n>
0;255;3;0;9;TSP:MSG:BC<\n>
0;255;3;0;9;TSP:MSG:FPAR REQ (sender=10)<\n>
0;255;3;0;9;TSP:CHKUPL:OK<\n>
0;255;3;0;9;TSP:MSG:GWL OK<\n>
0;255;3;0;9;!TSP:MSG:SEND 0-0-10-10 s=255,c=3,t=8,pt=1,l=1,sg=0,ft=0,st=fail:0<\n>
0;255;3;0;9;TSP:MSG:READ 10-10-255 s=255,c=3,t=7,pt=0,l=0,sg=0:

I can see that the gateway is receiving some messages from the node but it "fails?!" to respond accordingly?!

Anybody could enlighten me?

Gatewaycode = 99,99% vanilla except credentials and mqtt broker (mosquitto)

"light sensor" node code only for completenss, ignore the stuff that is commented out

#define MY_NODE_ID 10 <--- manual node id is set!

#include <Streaming.h>


/**
 * The MySensors Arduino library handles the wireless radio link and protocol
 * between your home built sensors/actuators and HA controller of choice.
 * The sensors forms a self healing radio network with optional repeaters. Each
 * repeater and gateway builds a routing tables in EEPROM which keeps track of the
 * network topology allowing messages to be routed to nodes.
 *
 * Created by Henrik Ekblad <henrik.ekblad@mysensors.org>
 * Copyright (C) 2013-2015 Sensnology AB
 * Full contributor list: https://github.com/mysensors/Arduino/graphs/contributors
 *
 * Documentation: http://www.mysensors.org
 * Support Forum: http://forum.mysensors.org
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * version 2 as published by the Free Software Foundation.
 *
 *******************************
 *
 * REVISION HISTORY
 * Version 1.0 - Henrik EKblad
 * 
 * DESCRIPTION
 * Example sketch showing how to measue light level using a LM393 photo-resistor 
 * http://www.mysensors.org/build/light
 */

#define MY_NODE_ID 10
#define MY_BAUD_RATE 9600

// Enable debug prints to serial monitor
#define MY_DEBUG 

// Enable and select radio type attached
#define MY_RADIO_NRF24
//#define MY_RADIO_RFM69

#include <SPI.h>
#include <MySensors.h>  

#define CHILD_ID_LIGHT 0
#define LIGHT_SENSOR_ANALOG_PIN A3

unsigned long SLEEP_TIME = 1000; // Sleep time between reads (in milliseconds)

MyMessage msg(CHILD_ID_LIGHT, V_LIGHT_LEVEL);
int lastLightLevel;

//#include "WatchdogAVR.h"
//typedef WatchdogAVR WatchdogType;
//WatchdogType Watchdog;

void setup()
{
  //Watchdog.disable();
  //int countdownMS = Watchdog.enable();
  
  //Serial.print("Enabled the watchdog with max countdown of ");
  //Serial.print(countdownMS, DEC);
  //Serial.println(" milliseconds!");
  //Serial.println();
  
  //Serial.begin(57600);
  Serial.println("setup() begin");
    // LightSensor
  pinMode(A3,INPUT_PULLUP);
  pinMode(A2,OUTPUT);
  digitalWrite(A2,LOW);
}
void presentation()  {
  // Send the sketch version information to the gateway and Controller
  sendSketchInfo("Light Sensor", "1.0");

  // Register all sensors to gateway (they will be created as child devices)
  present(CHILD_ID_LIGHT, S_LIGHT_LEVEL);
}

void loop()      
{     
  static long vcc = readVcc();
  static int vccpercent = map(vcc,2400,3007,0,100);
  sendBatteryLevel(max(min(vccpercent,100),0),false);
  Serial << "vcc: " << vcc << endl;
  // Required for ack
  //wait(100);
  
  analogRead(LIGHT_SENSOR_ANALOG_PIN);
  int lightLevel_raw = analogRead(LIGHT_SENSOR_ANALOG_PIN);
  int lightLevel = (1023-lightLevel_raw)/10.23; // as of 1023 !!
  lightLevel = 511;
  Serial.print(lightLevel);
  Serial.print(", ");
  Serial.println(lightLevel_raw);
  //if (lightLevel != lastLightLevel) {
      if(!send(msg.set(lightLevel),true))
        Serial.println("Send failed!");
      lastLightLevel = lightLevel;
  //}
  //Serial.print("RETR=");
  //Serial.println((0x0F & RF24_retrycount()));
  wait(100);
  sleep(SLEEP_TIME);
}
// https://forum.mysensors.org/topic/3463/m_ack_variable-or-m_set_variable/2
void receive(const MyMessage &message) {
  if (message.isAck()) {
      Serial.println("This is an ack from gateway");
      //Serial.println("Reset Watchdog!");
      //Watchdog.reset();
      }
}

I got headache from this line:

0;255;3;0;9;!TSP:MSG:SEND 0-0-10-10 s=255,c=3,t=8,pt=1,l=1,sg=0,ft=0,st=fail:0<\n>

After the node requsts a parent .. isn't the node supposed to listen for the response? Where should this happen in the code?!

[tekka]

@cimba007 I assume

wait(100)
sleep(SLEEP_TIME)

is too short to fully reconnect.

Try this code instead (this prevents the node from sleeping if transport is not operational)

if(isTransportOK()){
    wait(100);
    sleep(SLEEP_TIME);
  } 
  else {
    wait(SLEEP_TIME);
  }

[cimba007]

@tekka I will try it at once! I was thinking about something like this too. Thanks for pointing out that using isTransportOK() might be the solution.

A quick first test is looking very good. :relieved:

Is it possible to include this wait in the send-function? The self-healing feature of the network would be completly transparent to the enduser. Maybe add some configurable timeout function or another paremeter to send?

[sola]

I have the exact same problem: the node never reconnects to gateway after a gateway outage.

In my case it is an ethernet MQTTClientGateway and a humidity node.

I have experimented with your suggestion and found that 1000ms was still not enough for a full reconnect before the node goes to sleep, so I made it 3000ms. This seems to allow a reconnect all the time (at least in my environment).

  // @TODO remove after testing gateway outage
  if (!isTransportOK()) {
    #ifdef MY_DEBUG
    Serial.print("Transport ERROR. Waiting for a proper transport reconnect");
    #endif
    wait(3000);
  }

  // Sleep for a while to save energy
  sleep(UPDATE_INTERVAL); 

I am unsure why MySensors doesn't have this self-healing element built-in. Maybe it is only a bug, maybe it has some deeper reason.