2.3.5.152 upgrade and virtually everything is broken

I had tried multiple previous restores to old db's (past day) and didn't help/work. It took the restore back to 2.3.4.148 to get that P1 in particular back up and running.

I will continue to troubleshoot and I am working on new wifi. (within 2 weeks/ordered) I would like to know if anything is found in the code and whether I will be able to upgrade in future.

Hey big thanks rlithgow1,KK, aaiyar and all the other folks! Let me know if you want me to try anything else out.

Edit: Looks like it is network. Everything started great and now grinding to a halt again. Number of drop offs. I hope it is indeed network as investing in replacement is not cheap and I have spent a pile of time on this already.

1 Like

@goldbond1 while you troubleshoot your network setup, you can leave the experimental switch driver for your plugs.

I thought that my C-8 Zigbee network is nearly perfect, but obviously, it is not...
I have 0.6% of the pings lost :

1 Like

Last night I tried to change the channel once again. (went to 12)
This morning I woke up to this:


I have gone ahead and upgraded to 152.

Edit: Amazingly, many of the devices are being found and building out this time on 152. I was wrong that the build broke things. More likely related to whatever is knocking Zigbee network offline and/or the fix was apparently when KK asked me to reset one of the repeater plugs. So, it looks like many of the Aqara are recognizing channel change (which can be an issue) and 152 I have a network. This has been painful but hope I am getting somewhere..

KK how are you getting ping fails? I don't see that in my driver reports. I suspect because I am using the SONOFF instead of the DOGAIN?

I received the 2x TP Omada 653's today. @aaiyar hoping you can verify if this is what you were talking about:
.

  • 653 - I set 2.5ghz to channels 1 and 6. (20 MHZ width) Medium Tx Power
  • 653 - Wifi 5 is set to channel 44/5220mhz. (160Mhz) Medium Tx Power
  • HE - On build 152. I am moving HE Zigbee to channel 20. The Zigbee network is currently rebuilding.

I'd leave 5ghz on auto

Thanks @rlihgow1 ..Looks like @aaiyar and you were correct. Somehow that XE75 POS scanned and destroyed my network costing me countless hours. Devices are adding easily so far. My motions are still acting a little funky at times. (1 motion is reversing the call (turning off light instead of on) Almost all devices are reporting in with some prodding. (re-add)

I have 1 last problem can't solve and tried googling. After messing around with all the apps to get this going, I have a handful of apps claiming installed (checkbox) and they are not there under App Mgr.

I'll update again when the network has stabilized.

2 Likes

Sounds good to me. I have no experience with Omada access points, but people who have them seem to like them.

For the Decos, the consensus seems to be:

  1. Use something else as a router/firewall (eg. pfsense)
  2. Setup the Decos in access point mode
  3. Optimize the WiFi network once (it'll put them on channel 2 with 40 MHz width)
  4. Block the Decos from accessing the internet at the firewall

They will work fine and not interfere with a zigbee mesh that uses channel 20 or 25. The Decos won't get any firmware updates, and you will be unable to use the Deco app to manage them, but all those functions can be done at the firewall/router anyway.

1 Like

My XE75s were setup in AP mode. I have a business grade router. The XE75 app still works for configuration and allows for wifi channel scan. (at least it claims it does) Like you said, the 40 Mhz update seems to have been the cause of my problems. And, I cannot adjust the channels on the XE75.

I have been working on my network for almost 2 hours now and 90% of the Aqara's are firing routines as intended immediately so far. 1 seems to reverse periodically as mentioned.

It won't if you block the Decos from accessing the internet.

1 Like

I woke up this a.m. with about 1/3 of the devices knocked off again after re-adding yesterday. Doesn't appear related to Sonoff locations. Mix of Dogain and Sonoff.

I'll try new drivers and re-adding yet again..

What is the connection stability of your mains-powered Zigbee devices after the LAN changes?
Set up RM5 polling ping tests, if not done yet... Assign temporarily the Tuya Zigbee Switch driver to your Sonoff and Dogain plugs.

Here are my stats 3 days later :

  • Plug TRADFRI control outlet LivingRoom : pingsOK=17141, pingsFail=22 (0.13% packets loss)
  • Plug TZ3000_ps3dmato AC LivingRoom : pingsOK=13674, pingsFail=54 (0.39% packets loss)

The AC plug is located at the most distant location from my C-8 hub, 2 solid walls in between, so the the higher percentage of failed pings can be expected.

All of your Zigbee routers must have a very good connection to the hub first, before looking at any other devices.

1 Like

Sorry taking so long. I am trying to figure the rule logic and was out of town 1 day. I will try and get this done before end of weekend.

Well I can safely say just about everything is still broken...but new symptoms.

  • 2x EAP653s installed on opposite sides/floors of house with wifi 2.5Ghz channels 1 and 6 and HE is now on 22. Wifi5 is automatic for both. WPA2/AES used. Now the KASA's drop offline about every 15 min for about 30 seconds. Great..

  • Purchased more DOGAIN to replace the SONOFF hoping to stabilize repeaters or elimate a few remaining drops of devices. Can't even sync with HE. They get added with wrong driver and won't correct. They are about 12 ft from the HE when syncing. Never had this issue before.

  • The only good news is the HE seems to be 90% stable with devices. Still some stupid stuff going on.

Overall, this is a nightmare and pretty much ready to go back to the dark ages of using light switches..

Simply switch the driver, click save then configure (this is crucial)

I tried that several times. But, it does nothing. I am trying to switch to Generic Zigbee Outlet you guessed it OLL? How the heck did it default to that? Will keep trying

On the good news front, I am hoping swapping out my 802.3af POE+ switch with an 802.at solves the KASA problem. The 653 is supposed to be 15.4 watts and the EAP653 13.9 watts. But, I suspect there is something wonky there. All I know is I need 1 win tonight... - update still happening. Starting to wonder if 35 Kasa is just too much..

So this just in...
Am I correct that the wifi broadcast on this stupid Vizio soundbar (which won't disable) is going to conflict with HE channels 20 on up likely? (2.5 Ghz)

I also found out the KASA are flashing likely because I am blocking outbound internet. I turned on for 2 hours and no flashing.

Thye OLL driver is being automatically selected becuase it has the best fingerprint match to the device. I've been using many of those drivers with my Zigbee devices. You do have to be careful with the recovery method built into those drivers. The recovery method can & will take down your mesh if set too aggresively. My settings are "normal" and "1 minute", anything greater than those settings has affected my Zigbee mesh. Now several devices previously run by OLL drivers are now running much better on @kkossev 's drivers. I know Krassimir said that the drivers shouldn't affect the devices staying connected, but my experience is my Tuya USB Zigbee repeaters are 100x times more reliable with his driver, when prior they would constantly fall off with the OLL driver. I keep my Orbi Wifi Sats on Channel 1 & 20mHz.

1 Like

There is an option on those to turn off cloud binding on the device page, then they should not try to get out anymore and may stop complaining. I have it turned off on mine and I am not seeing any kasa related DNS requests. Only side effect is you cannot control them from the Kasa app remotely, or any other Kasa cloud integrations. I run everything on mine through HE so its not an issue.

image

1 Like

Last night I updated from 2.3.5.146 to 2.3.5.152.

After the upgrade:

  1. as of this morning, four of my zigbee devices are totally offline. They are not showing as being part of my zigbee mesh, and in the chart of zigbee devices, they show last message as "N/A" and the total message count is 0, which I think indicates they have not reconnected since the hub rebooted after the update.
  2. my nighttime routine did not fire last night, but I'm not yet certain why. It does not rely upon zigbee devices.

ETA:
a. My HE is a C7
b. The Zigbee devices that dropped off are various different types, including a Sonoff S31Lite (but another is working fine), a Sylvania bulb, an Aquara sensor and a Comcast door position switch.
c. My hub uses zigbee channel 20 and has for over a year
d. Four of my z-wave devices had not yet paired with the hub after restart due to update. I was able to get those to pair by toggling their on/off status.
e. Per one of the above posts, I backed up the database, and did another restore in case there was some sort of error when the hub restored after the update. That resulted in more zigbee devices dropping off, and different zwave devices dropping off.
f. Per one of the above posts, I tried to toggle the mains powered zigbee devices after the database restore. They are totally unresponsive, so I can't toggle them.
g. Per another of the above posts, I tried to power cycle the HE in case the update and/or restore don't power it down for long enough. I left it off for about 6 minutes. When I turned it back on, EVEN MORE zigbee devices have dropped off, and again a different set of zwave devices are dropped.

  • The zwave devices will connect by toggling them, so I suspect they would connect when a HE rule called for them to go on.
  • The zigbee devices will not connect by toggling from HE. I'm trying to pair the mains powered ones. After physically unplugging one and reinserting it, I got it to connect. As a result, some battery powered devices then immediately reconnected. I think physically unplugging the zigbee devices might be what is needed.

ETA 2: > Four hours after physically unplugging the powered zigbee devices (among the zigbee devices that had never reconnected >9 hours after the initial hub upgrade) and replugging them and them connecting, all the battery powered devices finally reconnected per the zigbee details matrix. They are not yet showing on the Zigbee graph, so I don't know if they connect through other devices, but they are working.