ESP32-C3 becomes unresponsible after a couple of days

[AdmiralRaccoon]

What happened?

I´ve already read that the ESP32-C3 is not the best solution, but maybe it´s a bug anyway causing the issue:

After a couple of days, the device because unresponsible within Home Assistant as well as pinging it / access to the WebUI. Even pressing the button for toggeling won´t work. I have to remove power to get it working again, and then it will work again for the next 3-5 days.

I will notice when it stops working in the morning, as it supposed to wake me up in the morning.

To Reproduce Bug

Waiting several days

Expected Behavior

Working 24/7

Install Method

Binary from WLED.me

What version of WLED?

WLED 0.15.0-b7

Which microcontroller/board are you seeing the problem on?

ESP32-C3

Relevant log/trace output

No response

Anything else?

It´s a WLED Slim LED Strip Controller from Athom:
https://de.athom.tech/blank-1/wled-slim-led-strip-controller

It´s doesn´t get hot
WIFI sleep is disabled
WIFI TX Power is at 8.5 dBm (that was the default)

Code of Conduct

• I agree to follow this project's Code of Conduct

[netmindz]

Is this an issue only after you upgraded to 0.15 ?

What version was it previously stable on?

[AdmiralRaccoon]

I used the 0.14.4 for about a week after i bought the controller, but had an issue with presets, which were actually fixed with the 0.15, so i directly started using 0.15.0-b6 (which already had these unresponsible issues)

[DedeHai]

This could be a hardware related issue as well: please try this: add a large 1000uF capacitor to both 5V input and 3.3V rail on the ESP and report if the issue persists.

[AdmiralRaccoon]

Now that exceeds my knowledge, as this is a ready-to-user controller with an usb-c power input:

I thought about that it also could be hardware related issue, but I rather replace the C3 with an ordinary ESP32, as these work rock solid with wled.

[KrX3D]

This could be a hardware related issue as well: please try this: add a large 1000uF capacitor to both 5V input and 3.3V rail on the ESP and report if the issue persists.

hi, thanks for that tip, i will try it on my devides since i got the same problem but not just with the C3 but also with s2 and some esp32 dev boards

[daggerrz]

I think this issue more likely is related to an underlying problem with the ESPAsyncWebServer dependency. @AdmiralRaccoon, are you using Home Assistant by any chance? HA can be pretty aggressive in its polling and open many concurrent conenctions to the WLED device, which might lock it up.

See me-no-dev/ESPAsyncWebServer#157 and me-no-dev/ESPAsyncWebServer#1116

This is still easy to reproduce with a synthetic load:

wrk -c 1 -d 20 -t 1 --timeout 5 http://IP/json

Increasing concurrency, -c to e.g 8 will consistently lock up an ESP32-C3 on 0.15. I haven't tested other ESP32 variants recently, but they will likely fail around the same level of concurrency as well.

[AdmiralRaccoon]

So might this be something for the HA WLED Integration Developer to look at and do something about polling interval?

[willmmiles]

I've been working on some upgrades to the web server components to better handle concurrency and excessive load in #4119 -- the underlying AsyncTCP library melts down rather dramatically if its event queue overflows. While the code there will probably improve load related problems for C3s, I've still got some unresolved issues in low memory environments such as S2s.

[mattfox27]

I have been having the same issue, and yes im using home assistant