HASPone appears to restart itself often

[drderiv]

I started noticing a quick flash every 3 to 5 minutes on my HASPone display. I set a camera on it and in looking frame by frame, it's super briefly showing the "WiFi Connected" screen, which says that both WiFi and MQTT are connected. Checking the HA logs, I get a sequence of "plate01 backlight became unavailable", "plate01 sensor changed to OFF", followed shortly by the same messages with on states. Everything seems to be working, and connection strength and logs of my UniFi network shows no issues there, with strong signal. Any suggestions on how to chase down what appears to be an issue with the HASPone frequently restarting?

[aderusha]

First thing you'd probably want to know is if it's actually restarting or just loosing wifi. If you don't see the "HASP Initializing" screen, that might suggest it's dropping connection and reconnecting for some reason, which itself is almost always radio related.

[drderiv]

Got it, and from looking at a higher speed/resolution video, the Initializing screen never comes on. Perhaps interestingly, when the "WiFi Connected!" screen first appears, it also says "MQTT Connected:" but then a couple frames later, it switches to "MQTT Connecting:". So I guess this seems like it's being triggered at the top of the loop()? The strange thing is, there's nothing about a disconnection in my wifi logs, and it UniFi shows the wifi experience of the HASPone to be 100. When my phone disconnects or even changes access points, there's a log entry. Is it possible that only MQTT is getting dropped and not wifi? I'll also note that the device can be sluggish, meaning it will ignore screen touches much of the time, like it's busy doing something else.

[aderusha]

It might be useful (if annoying) to take the device out of the wall, pull it apart, and hook to the USB port buried inside (it'll power up via USB so you won't want AC connected). Then you can monitor debug logs even if WiFi isn't working, and see if you can't figure out timing or if any errors are showing up. The device itself does nearly nothing, there are no background tasks or anything like that, it's purely a gateway between incoming MQTT messages which have Nextion commands inside, it'll then send the Nextion command to the screen. In my experience, any latency or delay etc is essentially always down to a network/wifi issue and you can validate that by monitoring timestamps in the log. Sadly, the ESP8266 RF and wifi performance isn't always great.

[drderiv]

Thanks for the advice. I've done what you suggested and it looks to me like the wifi never disconnects, as suspected, as I never get the "Reconnecting to WiFi network..." message in the serial output. I do, however, get the "MQTT: not connected, connecting." message regularly. This snip of the log file has one cycle of that. My HA instance is running on a Windows server that has no known issues and is continuously available to me via the HA app or local website. I'd also curiously note the occasional garbage character in the output which makes me suspicious that I may have a bad ESP8266? If you have any other test or debugging suggestions, I'm happy to try them.

[aderusha]

What I'd want to track down is correlation between timestamps on HASPone logging versus actions taken in Home Assistant. As you note, HA performance is probably not in scope here as it should be doing nothing more than firing off an MQTT message. The test is to see how long it takes between that message being sent and the HASPone acknowledging its receipt. If the delay is substantial, it points to wifi problems which leaves a lot of further troubleshooting to do.

[drderiv]

Okay, so how to do that? And what could explain the garbage characters coming from the D1 that is connected via serial?

[aderusha]

Annoyingly, step 1 is to disassemble the unit as you'll need access to the USB (while logging is visible via telnet, any delays in the network would be masked as the logging is happening over the same network). With that connected to a PC with a terminal (and disconnected from AC power), you should be able to view the log while you interact and/or send commands from the HA side.

Garbage characters over the serial log might not be a problem, there are technically two hardware UARTs on the D1 but for reasons only one is usable, so that is used for communications with the Nextion. The serial debug messages are sent via SoftSerial which can be buggy, particularly when run at high speed. What I'd be more worried about would be errors from the Nextion complaining about malformed commands and those you'd see in the logging. In either case, you can adjust serial speed downward on the web admin page which should resolve any issues there. The only real impact the higher speed settings will make is to improve performance when flashing a new firmware to the Nextion, which most people will never need to do.

[drderiv]

I have that all set up, which is of course how I created the log file above. So to test this, you want me to trigger actions from the HA side, then see how long it takes for messages generated by the actions to show up in the D1 serial output, correct?

[aderusha]

Exactly that. If there is a delay, that strongly points at problems with wireless. The ESP8266 is single threaded and there are no background tasks etc running (at a high level anyway). These transactions should be perceptively instantaneous, you send a command from HA and you should see a log entry within a few milliseconds. If that isn't happening, something in the wireless world is having trouble.