Connection problems with ZHA

Heyhey
Im using the Home Assistant Skyconnect with Firmware 7.4.2.0.
In my ZHA Zigbee network there are 95 Devices.
There are 33 router devices
24x thirdreallity plug
1x third reallity nicht lieght
1x Phillips Hue Plug
2x Aqara Plugs
2x Aqara LED Strip
1x Innr bulb

The other devices are battery powered devices -
many Aqara HT
many frient smoke detectors
some Aqara buttons
some thirdreallity buttons
some Aqara water leak detectors
two magnetic water valves

Most of these devices are workin perfectly fine.
But I sometimes turning something on and off doesnt work an I get the error message:
"Failed to send request: Failed to deliver message: <sl_Status.ZIGBEE-DELIVERY-FAILED-3074>
Trying it three or four times usually works. (The device succesfully turns on or off than).

The (mains powered) devices causing this problem are always the same.
Everything battery powered works as expected (or I dont realize the communication problems)
If I repair the device in question and turn it on/off directly after its new connected it works instantly.
I click on turn on and its on without any noticable delay.
If I wait some hours the problems start to begin again.

I really wonder why this happens. I exclude connection problems, because of all that router devices and the fact that If I rejoin the device to my ZHA network it works instantly.

The Skyconnect is plugged to an USB 3 port (because my intel nuc doesnt have any USB 2 port), but Im using a 3m long USB 2 extension cable which is glued to my wall (about 2 - 2.5m away from my nuc) and I plugged a (not used) third reallity plug about 2m away from my Skyconnect.
So I think this is fine aswell.

Im wondering if my change from Sonsoff Zigbee Dongle E to the skyconnect (because I wanted to support the Nabu Casa) was a good idea.
Is there any answer to the question which dongle is better? (Skyconnect or Sonoff)

And is there anything I can do to solve my zigbee issues?
Im on and on to change the questionable devices to Shellys, but removing Zigbee routers wont make it any better ofcourse.

If it does matter. In addition to zigbee Im also using some HmIp (433Mhz) and Shelly (WLAN).
Those two are working perfectly fine (with some very very rare connection problems.
And I really dont think that shelly plugs are disturbing the zigbee network, because in some cases they are really really far away.

I also added a Screenshot of my zigbee mesh.
I think this really looks like a mesh network should look like.

FWIW, i got that specific error for the first time today when an animation tried to shut down a simple switch. The only new thing I added to my environment recently was a z-wave dongle for a siren. Unsure if that had anything to do with it

Best practice is to start by reading and try following all the tips here before troubleshooting any deeper → Zigbee networks: how to guide for avoiding interference + optimize using Zigbee Router devices (repeaters/extenders) to get a stable mesh network with best possible range and coverage

It could be that you have one or two devices that are faulty or have buggy firmware, but with those symptoms the most common problems is either caused by EMI/RMI/EMF interference or by the user simply just not having enough “good” Zigbee Router devices, as well and placing those “good” Zigbee Router devices in strategic locations to spread them out to get best possible range and coverage

Suggest that you buy many more Zigbee Router devices or alternatively a few “known good” (or better make yourself a couple “known great”) dedicated Zigbee Router devices. First though, before purchasing anything I highly recommend you begin by reading my whole guide and try to follow all the best practice tips there → Zigbee networks: how to guide for avoiding interference + optimize using Zigbee Router devices (repeaters/extenders) to get a stable mesh network with best possible range and coverage

Zigbee is both susceptible to EMI/RMI/EMF interference and more importantly Zigbee relies heavily on mesh networking so it depends on you having many Zigbee Router devices as part of its mesh network.

That linked guide includes some specific products that are good or great Zigbee Router devices, and it also include the tip to try changing to a different Zigbee channel if nothing else is helping/working (which is now even possible to change in the ZHA UI to make it easier) → https://www.home-assistant.io/integrations/zha#defining-zigbee-channel-to-use

Otherwise you would normally only need to keep adding many more Zigbee Router devices until your mesh network is built out to active good reception to the required range and coverage.

While using a USB 3.0 port might be unlikely to be the root cause for your issues it is still known to cause issues so better remote that factor if possible as USB 3.0 (really all USB 3.x and USB 3.x) is infamously known to cause serious EMI/RMI/EMF interference issues with Zigbee (as well as Bluetooth and Thread) in some circumstances, so recommend you just to be sure buy a powered USB 2.0 Hub (a USB 2.0 Hub with its own external power-supply) and connect your Zigbee Coordinator adapter dongle via it instead. That USB 2.0 hub will in practice act like a USB 3.x to USB 2.0 converter and as such at least isolate the Zigbee Coordinator adapter dongle (and can also be used with ). Such USB2 hubs usually cost less than $20(USB) and are totally worth it to eliminate that potential issue, see example → Amazon.com : usb 2.0 hub powered

Again, I highly recommend you read my whole guide regardless which among many other best practice tips also include links to and empirically documented tests that show how USB 3.0 will interfere with Zigbee signals → Zigbee networks: how to guide for avoiding interference + optimize using Zigbee Router devices (repeaters/extenders) to get a stable mesh network with best possible range and coverage

That should not be any huge difference (as long as using a long USB extension cable to a USB 2.0 port to avoid EMI/RMI/EMF interference), becuase Zigbee relies so heavily on mesh networking so it is really all your Zigbee Router devices that should make the mesh network work, with almost just the closest Zigbee Router devices connecting directly to the Zigbee Coordinator and all the rest only connecting indirectly via other Zigbee Router devices.

However, note that EmberZNet 7.2.x.x and later firmware is yet mature enough to be well proven as stable (though the EmberZNet 7.3.x.x has so far tested very good), thus it is probably a good idea to downgrade for now, and do make sure to use the EmberZNet “NCP” (i.e. the dedicated Zigbee firmware) and not the “RCP” Multi-protocol/Multi-PAN firmware for Zigbee and Thread (as Mult-protocol/MultiPAN firmware for Zigbee is no longer maintained because it is known to have issues and not be stable Multi-protocol was always only experimental anyway), check out:

and

Forgot to ask what firmware type and version you are using on your SkyConnect (ZBT-T) USB dongle?

Oh, and in case you missed that tip in the guide, your old Sonsoff ZBDongle-E adapter is actually the product that makes the best known dedicated Zigbee Router device if you flash it with Zigbee Router firmware and then simply power it with an USB-charger to make a “known great” stand-alone Zigbee Router (repeater/extender) → ITead’s “Sonoff Zigbee 3.0 USB Dongle Plus V2” (model "ZBDongle-E") based on Silicon Labs EFR32MG21 radio SoC/MCU

See this concept example for using Sonsoff ZBDongle-E adapter as a stand-alone Zigbee Router:

Hey @Hedda
Thank you very, very much for your detailed posts.
Ill take a deep look into this at the weekend - as I suppose it will take some time to read (and read again I suppose) and understand.
Ill give feedback at the beginning of next week.

By the way, Zigbee devices from Aqara (and Xiaomi) are infamous for having buggy firmware that known issues connecting to closer/better Zigbee Router devices unless you manually re-pair each and every Aqara/Xiaomi device after you added additional Zigbee Router device as well as being very fussy about what brands of Zigbee Router devices they work well with. See many more Aqara (and Xiaomi) specific tips and details here:

So not only do you need to avoid EMI/RMI/EMF interference and have enough “known good” Zigbee Router devices, you need to add/have Zigbee Router devices that are known to be compatible with Aqara (and Xiaomi) and then you also need to re-pair Aqara (and Xiaomi) devices if you add any more Zigbee Router devices.

And of course also never ever power-off Zigbee Router devices that Aqara (and Xiaomi) connects to.

@TheSmurf More or less when you noticed those errors, so did I, and I am pretty sure they were newly occurring. My mesh is strong, balanced (routers vs end devices), and I only employ devices from good brands (those that as far as I can tell are diligent in following the Zigbee 3.0 standard). I have also thoroughly investigated interference and taken the necessary steps to mitigate its effects. That said, the issue popped up along with previously very reliable suddenly sensors not working. If I recall correctly there was a recent ZHA update which may or may not be related.

Also, it appears others are reporting the same or similar issues here:

However, when the logs of the person who opened the issue were reviewed the analysis was that Tuya mmWave and metering plugs were flooding the mesh. Tuya is one of the brands I avoid due to many reports of issues attributed to unruly devices, lack of full compliance, etc.

On my end, I had a dozen or more Samjin (rebranded Samsung SmartThings from my prior 7yrs on ST) plug in outlets that I removed to alleviate traffic on my mesh (~90 devices) and things were better after doing so. I swapped most of them, along with some Jasco in-wall outlets, with ~20 s31 outlets running on wifi (ESPHome).

I removed the Jasco outlets because I had a strong impression that whenever I had power issues during storms, one or more of those Jasco outlets would misbehave by freezing up routing for all the devices in their vicinity. All those outlets were providing most of the routers for my zigbee mesh. Now that they are gone, I needed to add new routers so I replaced ~12 Zwave ceiling and exhaust fan switches with Inovelli Zigbee equivalents (love them!) resulting in a robust mesh that was working great prior to this error popping up.

To provide some background… given most rooms either have a ceiling fan or an exhaust fan along with a light, I decided to keep using zwave for the light, and replace the zwave switches for the c/e fans with zigbee equivalents to ensure a well distributed zigbee mesh. There are over 40 routers but many are clustered and not wired into the house (so can be removed causing issues) therefore these 12 routers should help increase reliability.

Why all of this info? There is a lot of focus on interference, rightfully so, and not enough on proper engineering of the mesh. A few other aspects I feel are equally important are:

  1. Product choice - look for good quality products from reputable manufacturers that are committed to adopting the latest standard and above all complying with it diligently. When picking a manufacturer, also check whether they make firmware updates available as that allows them to fix bugs and improve compliance.
  2. Mesh build out - there are a lot of best practices here, but boiling it down to basics, one could say that the routers chosen (what product), their number, and location are basic elements that can often be overlooked. In other words, one might just focus on the cool new sensor they want to add, and ignore the possible need for a strategically placed router (even if not needed for its main purpose) needed to ensure the sensor works properly. Even if the sensor appears to work without the close-by router, it might drain its battery quickly if it struggles to communicate with the coordinator (repeated messages, higher power output if applicable, drain the battery faster).

I am probably only scratching the surface, but this post is becoming way too long. Hope it helps and/or inspires good mesh choices :slight_smile:

1 Like

I use the SkyConnect Firmware 7.4.2.0.

You need to know that the network is within my store (about 1000m2 plus 2 storeys with office and stuff). So its kinda big area and few (usable) power outlets. Because of metal shelves infront of it).

First of all, I did replace my Innr bulb with a shelly bulb, because I really need this guy to work. (Not turning off this light - caus of zigbee problems - causes trouble here)

So, after reading all that linked stuff I really dont believe in interferance problems of my coordinator. (But ofcourse Im open for arguments against my thoughts)
Its always the same devices not working sometimes.
If I turn on my lights its always the same lights I have to look if its turned on/off.
At the outside its always the same plug making problems.
If it would be coordinator interferance, it should be random devices not working, shouldnt it?
And I plugged my coordinator in about 2 - 2.5m distance to my homeassistant with a shielded (amazon basics) USB2 cable.
And the plug directly next to my coordinator (which has the only purpose of strenghten the mesh, but should be connected directly to the coordinator as its about 2m away) always turns on and off without any delay if I test it.

So, what I would do next (please tell me if im on the right way):

  1. Replacing my 2 Aqara plugs with 3rd reallity plugs.
  2. Downgrade my Skyconnect firmware to 7.2.x.x
  3. Flashing my Sonoff Dongle with this zigbee repeater firmware and get it in the middle of my shop (where there is no zigbee device so far, because I cant use the power outlets there, caus of the shelves directly infront of it that will kill my signals) and put it on top of this shelf - with a long USB cable)
  4. Getting some of those ikea repeaters putting it on top of my shelfes, too (as the sonoff dongle). Or getting more Sonoff Dongles instead, if they are way better.
  5. (Not sure with this)
    I have several Shellyplugs (about 12) very close to each other. One of those I replaced with zigbee plug. This plug doesnt work that reliable (often takes 3 or more trys to turn it on).
    So I think putting one zigbee device directly into 10 wifi plugs wasnt my best idea :slight_smile:

And my thoughts to the 3rd reallity hardware.
Ivechosen them because they got th “works with homeassistant” brand.
Im a bit confused that they need a custom quirk in zha, what should tell me they dont follow zigbee standarts.
But anyway I think they work well, because the ones that have good connection in my network are working perfectly fine.
And they offer a firmware upgrwde ability.
And ofcourse I already own them. So getting some more wouldnt increase the amount of hardwaremanufacturers in my network (what seems to be a problem sometimes, too)
Please tell me if this is wrong and I shouldnt use them.

Edit: Can anyone provide a tutorial Flashing my Sonoff dongle E with router firmware on windows?
Im just confused by the link provided here, so I tried to flash it with the Skyconnect Flash Tool. But this didnt find my dongle.

Editedit: I found a guide.
Si Ive downgraded my Skyconnect to firmware 7.2.0.0 and I added one Sonoff Zigbee Dongle Repeater to a place where there is no zigbee device around. Lets see.
How long does it take until the devices (the not aqara devices) realize there is a new router to connect with?

This is good advice. The USB3 interference comes from the RF energy generated by the USB3 data rate, so put the USB2 hub close to the NUC and use a longer USB2 cable to the controller. (But your network map doesn’t present any interference issues).

The only in-wall switches that have failed here were Jasco/GE.

My Zigbee network grew like rabbits. The network backbone is 15 of the Third Realty switches (mostly acting as routers). No Zigbee device is more than 2 to 3 meters from a router.

3 Likes

Hey, for what it’s worth I also observed the same issues for the first time in the last few days. No new devices etc, but I think it coincides with updating to HAOS 13.1 - is it the same for you??

Nope.
It defenetely has nothing to deal with HAOS 13.1 as I did not update so far.
Im still on 13.0

All of my Zigbee devices work just fine in HAOS 13.1. Identifying a source of interference is extremely difficult, usually involving disconnecting the possible source and retesting. (Or a sledgehammer whacking power poles to find loose insulators. Something that old hams would understand).

I also have this problem which started yesterday. I have not made any network changes. It seems to have fixed away with the 2024.8.3 build I just installed?

I was constantly getting “<sl_Status.ZIGBEE-DELIVERY-FAILED-3074>” on all myHue devices linked to ZHA.

There is enough independent people reporting this error over the last few days that people should look beyond the typical device incompatibility.

I have no other insight, but I am hopeful the fix that showed up for me in 2024.8.3 is not just coincidence. Earlier today I did full HAOS reboots which didn’t help, but somehow this build did.

I can also confirm that this issue showed up in 2024.8.X (I can not pinpoint which patch release it was). My Zigbee network didn’t change and was really robust previously.

When I updated to 2024.8.3 my ZHA couldn’t finish initializing and all devices refused to connect. Also, it could not find new zigbee devices when I tried to pair a tradfri bulb. I noticed that they all dropped at the time of the upgrade, reverted to 8.2 and everything showed back up on the network, and I was able to pair the bulb.

My gateway is a SkyConnect, and I’m running HAOS 13.1.

Chiming in here with a link to the github issue related to this:

Was not getting this issue before 2024.8.3 tried to rollback to 2024.8.1 but am now still getting the same problem.

As mentioned there in that issue by zha devs, that is not a ZHA application-level wrror but instead a low-level error that comes from the Zigbee Coordinator stack which manages your Zigbee network the root cause must be enviromental, meaning interference or other which is best resolved by following best practices which includes taking actions to avoid interference and also adding many more Zigbee Router devices to strengthen your overall Zigbee network mesh (because Zigbee is very sensitive to interference and depend/relies heavily on mesh networking so need loads of Zigbee Routers) → Zigbee networks: how to guide for avoiding interference + optimize using Zigbee Router devices (repeaters/extenders) to get a stable mesh network with best possible range and coverage

I haven’t had this specific issue pointed out for the past year+ until I upgraded this week to 2024.8.X.

Makes me doubt it’s an environmental issue when others are suddenly seeing the same thing suddenly.

Unless something was being passed as a warning that is now erroring out, or potentially less lenient on passing data along.

FWIW, I downgraded to 2024.7.4 and it works fine now. Something is up.

1 Like

Now that did change as ZHA integration split out the Zigbee Gateway component into an external zha library in 2024.8.0 release where more errors are floated that did not do so before, see:

The longer back-story anout that change is summerized here:

So might be that you always had that problem but it was hidden in earlier versions.

I can confirm this.
Before 204.8.0 I had some zigbee devices sometimes just not working.
I feel the problems were less before, but thats only a feeling.

So I think Im having connection issues from my zigbee day 1, but I didnt get any error on that.
Now that I spend some time in improving my network I can see that the amount of errors (and not working devices) is getting less and less. And this in HA version 2024.8.3 here.