search

openshwprojects / OpenBK7231T_App

Open source firmware (Tasmota/Esphome replacement) for BK7231T, BK7231N, BL2028N, T34, XR809, W800/W801, W600/W601, BL602 and LN882H
https://openbekeniot.github.io/webapp/devicesList.html
1.5k stars 282 forks source link

OpenBK doesn’t reconnect after router reboots #509

Open ferbulous opened 2 years ago

ferbulous commented 2 years ago

Describe the bug I have a daily reboot schedule reboot for my wifi router at 5am. When I’m checking homeassistant, all the openbk devices are offline since 5am and I had to power cycle to get them connected again. Is there something that I need to enable in the settings to force reconnect to wifi?

Firmware:

To Reproduce Steps to reproduce the behavior:

  1. Reboot wifi router
  2. Openbk still doesn’t reconnect after router is online
  3. Power cycle required to reconnect to wifi

Screenshots If applicable, add screenshots to help explain your problem.

Additional context Add any other context about the problem here.

openshwprojects commented 2 years ago

It should reconnect by default. @btsimonh / @valeklubomir , thoughts? For me it reconnects, on my 15+ devices and i rebooted MANY times. image But I admit I had somehow a similar of issue in the past with my old crappy router. The solution was to create "ping watchdog". image

It was a long time since I needed that, can you check if it works?

ferbulous commented 2 years ago

@openshwprojects for my lights it worked, they do reconnect once the wifi is online not quite instantly for my wall switch though, it did came online after like 20 minutes since the router was rebooted maybe I need to extend the timeout longer?

openshwprojects commented 2 years ago

@ferbulous making timeout longer for ping watchdog will not make it reconnect faster. If anything, it will make it reconnect slower.

Are you sure that it takes about 20 minutes for that device to reconnect?

Wait, maybe you have the same issue as me a month ago and just your router can't keep up with so many devices?

I had a problem once where devices didn't want to connect to router and it turned out that my old TP link can't handle too many clients at once. I replaced my router with newer model and it helped.

Btw @ferbulous how are device groups? You have that black light issue, right?

ferbulous commented 2 years ago

Are you sure that it takes about 20 minutes for that device to reconnect?

Based on the homeassistant history logs

Wait, maybe you have the same issue as me a month ago and just your router can't keep up with so many devices?

Not sure if router related since I do have mesh wifi setup, I did earlier try flashing esphome (libretuya) on the same device and it reconnect just fine after router reboot.

Btw @ferbulous how are device groups? You have that black light issue, right?

The black color/light is still there on latest update. I'll add more comments on #503

WiredLife commented 1 year ago

Hey, i can confirm this bug, looks like a race condition. I reboot my two Belkin RT3200 with newest OpenWRT on it, then minimum of 1 from 9 lightbulbs doesnt reconnect, so i need to powercycle it. Sometimes 2 or more. Maybe it has something to do with 802.11r (roaming)? Like it can not chosse which channel to connect? I dont know

ferbulous commented 1 year ago

Hey, i can confirm this bug, looks like a race condition. I reboot my two Belkin RT3200 with newest OpenWRT on it, then minimum of 1 from 9 lightbulbs doesnt reconnect, so i need to powercycle it. Sometimes 2 or more. Maybe it has something to do with 802.11r (roaming)? Like it can not chosse which channel to connect? I dont know

Does it also happen randomly and goes offline for long period of time? This is my temporary fix for now using script, not the best solution if your mqtt broker crashed

//if mqtt offline, create timer eventid ‘01’ for 2 minutes and restarts
addEventHandler MQTTState 0 addRepeatingEventID 01 120 1 restart
//if mqtt online, cancels eventid ‘01’
addEventHandler MQTTState 1 cancelRepeatingEvent 01
openshwprojects commented 1 year ago

That's a very nice script and I can confirm that it should indeed work. I will just note that you don't have to do "01" index, it's simply 1.

@btsimonh do you know everything about this issue? I am checking my devices but all are up...

btsimonh commented 1 year ago

I don't reboot my router often. But I have booted my N device from the router to cause reconnect on occasion.
I have concerns about DHCP/DNS - same LWIP threading issue as MQTT. Also, MQTT 'gethostbyname' stalls the timer thread for a few seconds - I have a mod on the way which polls for resolution rather than wait. If may be a good test to specify the MQTT server by IP - this would avoid the 'gethostbyname' wait. @ferbulous - if you get a chance to setup as IP based mqtt config and retry, that may be useful diagnosis.