Fix issue #7 (MQTT reconnection)

[HamzaHajeir]

Hi

Fix to #7 where this is the scenario : = Router reboots = MQTT senses disconnection, onDisconnect called :

• MQTT Destroys client.
• H4P reschedules MQTT reconnect by this code Immediately : if(autorestart && WiFi.status()==WL_CONNECTED) h4.every(H4MQ_RETRY,[this](){ connect(); },nullptr,H4P_TRID_MQRC,true); = after H4MQ_RETRY (10 seconds) , it calls connect() :
• PANGO creates a "dead" client = ESP8266 lostIP event. = every H4MQ_RETRY, it calls connect() which ignores connection if there's a created instance of PANGO::TCP.

Here I've managed to (delay) the re-connection So network event can occur before trying to reconnect to MQTT. https://github.com/HamzaHajeir/h4plugins/commit/3e0b6236c7f0287ed969fa2b95b705edbdf188a0

This pull request contains another commit that solves another problem :

The problem is : After MQTT is established again after a reboot, and after 30 seconds of unreachable MQTT server : onDisconnect is called, in which it doesn't activate the reconnection lines because _discoDone = true.

That's because It doesn't pass through onConnect() callback function that sets _discoDone to false.

Removing _discoDone codes in H4P_AsyncMQTT solves the issue. And MQTT reconnects all the time without any issue.

Hope it helps

Hamza Hajeir

[HamzaHajeir]

Found bugs at this state, Closing until it's fixed.

[HamzaHajeir]

Hi

I've fixed and tested the code in my environment.

All nodes are responding well between router reboots and after long-time missing broker (on My PC OFF).