[Bug]: MQTT enabled with custom server/auth causes a reboot loop

[acidvegas]

Category

WiFi

Hardware

T-Deck

Firmware Version

2.3.6.7

Description

Enabling wifi and mqtt with a custom mqtt server that uses auth and tls is causing the device to get stuck in a reboot loop.

Relevant log output

No response

[oherrala]

@acidvegas Can you observe same things as documented in issue #3496 ?

[acidvegas]

@oherrala not I cant observe anything. as soon as I set a custom mqtt server with tls and auth, it just loop reboots over and over

[oherrala]

@acidvegas I don't know if there's documentation about debugging this stuff. I've been trying to figure these on my own. A quick guide:

Set device.debug_log_enabled to true (e.g. mestastic --set device.debug_log_enabled true) or from App.

After that you can open serial connection to the device via USB (e.g. screen /dev/tty.usb<something>) or try using the serial monitor on https://flasher.meshtastic.org/.

It should give you a log of what device is doing before it reboots.

[acidvegas]

@acidvegas I don't know if there's documentation about debugging this stuff. I've been trying to figure these on my own. A quick guide:

Set device.debug_log_enabled to true (e.g. mestastic --set device.debug_log_enabled true) or from App.

After that you can open serial connection to the device via USB (e.g. screen /dev/tty.usb<something>) or try using the serial monitor on https://flasher.meshtastic.org/.

It should give you a log of what device is doing before it reboots.

The logs from the serial monitor stop at:
INFO | 01:11:47 25 [WebServerThread] Attempting to connect directly to MQTT server mqtt.redacted.com, port: 1883, username: acidvegas, password: redacted

So I really can't debug past this point :(

Here is the relevant line of code for that log line, so it must be around here somewhere...

firmware/src/mqtt/MQTT.cpp

Lines 330 to 354 in 70712d8

	LOG_INFO("Attempting to connect directly to MQTT server %s, port: %d, username: %s, password: %s\n", serverAddr,
	serverPort, mqttUsername, mqttPassword);
	auto myStatus = (statusTopic + owner.id);
	bool connected = pubSub.connect(owner.id, mqttUsername, mqttPassword, myStatus.c_str(), 1, true, "offline");
	if (connected) {
	LOG_INFO("MQTT connected\n");
	enabled = true; // Start running background process again
	runASAP = true;
	reconnectCount = 0;
	publishStatus();
	sendSubscriptions();
	} else {
	#if HAS_WIFI && !defined(ARCH_PORTDUINO)
	reconnectCount++;
	LOG_ERROR("Failed to contact MQTT server directly (%d/%d)...\n", reconnectCount, reconnectMax);
	if (reconnectCount >= reconnectMax) {
	needReconnect = true;
	wifiReconnect->setIntervalFromNow(0);
	reconnectCount = 0;
	}
	#endif
	}
	#endif

My mosquito MQTT server is throwing:

714932188: Bad socket read/write on client <unknown>: Invalid arguments provided.
1714932225: New connection from REDACTED_IP:59348 on port 1883.

This is from testing without TLS (cause I didn't know if this was a TLS issue or not...)

[acidvegas]

I should also mention the T3-S3's and Heltecs are having this same issue. So it's not even a T-deck related issue.

Is anyone using a custom MQTT server with TLS & auth successfully?

ALSO, you can fix your device by running meshtastic --set mqtt.enabled false RIGHT after it boots. This will stop the reboot loop so you don't have to reflash the device to make it stop.

[acidvegas]

I have gotten a connect to my MQTT server (finally, was an ACL issue in moquito), but TLS causes a reboot loop still