@last_ping_request and @last_ping_response

[crawler]

Hi,
Found this issue when i was writing code that recovers broken MQTT connection.
Think that both @last_ping_request and @last_ping_response variables should be reset in #connect method, just before starting @read_thread

Here the code to reproduce it:

mqtt = MQTT::Client.new('mqtt://test.mosquitto.org')
sleep 24
mqtt.connect
mqtt.subscribe('#')

mqtt.get do |topic, payload| 
  puts topic, payload
end # => No Ping Response received for 23 seconds (MQTT::ProtocolException)
# In first keep_alive!

[ninp0]

To add to what @crawler is saying if you set:

mqtt.keep_alive = 0

It will immediately return from

ruby-mqtt/lib/mqtt/client.rb

Line 533 in b2e48d0

return unless @keep_alive > 0 && connected?

thus bypassing @last_ping_request / @last_ping_response evaluation.

That said, then it pretty much immediately begins to raise this error:

ruby-mqtt/lib/mqtt/packet.rb

Line 223 in b2e48d0

raise ProtocolException, 'Failed to read byte from socket' if byte.nil?

Here's the full stacktrace:

mqtt-0.6.0/lib/mqtt/packet.rb:223:in 'MQTT::Packet.read_byte': Failed to read byte from socket (MQTT::ProtocolException)
        from /usr/local/rvm/gems/ruby-3.4.1@global/gems/mqtt-0.6.0/lib/mqtt/packet.rb:30:in 'MQTT::Packet.read'
        from /usr/local/rvm/gems/ruby-3.4.1@global/gems/mqtt-0.6.0/lib/mqtt/client.rb:471:in 'MQTT::Client#receive_packet'
        from /usr/local/rvm/gems/ruby-3.4.1@global/gems/mqtt-0.6.0/lib/mqtt/client.rb:278:in 'block in MQTT::Client#connect'

UPDATE:

This appears to occur when an MQTT server is under a heavy load - I setup a local MQTT server and the problem is almost non-existent (i.e. it's observed much less frequently). Hoping there's additional resilience that can be implemented in the code to address this issue. Fwiw, this occurs quite a bit on meshdev:large4cats@mqtt.meshtastic.or:1883. More details to connect here: https://meshtastic.org/docs/configuration/module/mqtt/