Failed to deliver message: <EmberStatus.DELIVERY_FAILED: 102>

I’ve been getting the same errors for the last few months at this point. Linux desktop using VirtualBox for HA. I was considering switching to kvm but it looks like this is a cross-platform issue. I’m using a HUSBZB-1. I got a Skyconnect because I thought maybe that was the issue, but it looks like people are having the same problem with those, so I haven’t tried the Skyconnect yet.

I previously had issues with interference, and I never got an error message for that - things just didn’t work. A while back (well before this issue started), I redid my wifi and Zigbee to not be on the same frequencies as much as possible, and for some time things were working perfectly every time. Then this issue started out of the blue.

For me, nothing seems to clear up the issue except time - after waiting a while (a few hours, maybe a day or so), things will start working again. Rebooting the host machine, rebooting the HA VM, etc., none of that seems to make a difference for me.

Edit: I saw someone else was dealing with a busy frequency. Here’s my network scan. I’m using channel 24:

    "energy_scan": {
      "11": 28.30261646762903,
      "12": 28.30261646762903,
      "13": 21.09014924761344,
      "14": 10.914542804728702,
      "15": 17.086630587133605,
      "16": 2.2107128772756957,
      "17": 4.15070068297423,
      "18": 2.84844209578687,
      "19": 0.792717332355823,
      "20": 0.792717332355823,
      "21": 0.9017765778954641,
      "22": 0.6123372955913717,
      "23": 4.69985354430736,
      "24": 0.9017765778954641,
      "25": 0.792717332355823,
      "26": 17.086630587133605
    },

So that doesn’t appear to be my issue.

Edit 2 (really, edit #3, but I didn’t mark the first one): I moved my USB device away from the monitor (it was already on a cable away from the computer, but it was draped behind my monitor) and the issue cleared up immediately. So, maybe that was it, or it might have been just coincidence. Well, we’ll see if it comes back.

Edit 3: Yep it’s still happening.

UPDATE:

For me, the problem is gone. What I did:
Moved from Zigbee channel 20 to 22 and added a 1m usb extension cable.
Unfortunately, I can’t tell which one helped.

I have a total of 22 ZigBee devices at home plus a Sonoff gateway. HA is based on RaspberryPi 4. The devices are different (sockets with an energy meter, buttons, door opening sensors and IKEA Vinsdryka - temperature, humidity and PM sensor).
Everything was working until a week ago. HA I have updates all the time. So I wouldn’t associate the problem with switching HA to 2023.8, because I already had it two months ago. Anyway, the IKEA sensor started disappearing from the network. And only this sensor, the rest was visible. It’s not battery powered, so that wasn’t an issue here. Later I had problems pairing it with the network again. It took several hours before it could be paired again. I haven’t tried resetting the HA, I always deleted the sensor and tried to pair it again. Maybe it was a mistake.
Today I came across this topic because I managed to read this error using the ZHA toolkit. I changed the channel settings from Auto (15 was selected) to 25. We’ll see if it helps. For now, all devices except door opening sensors are visible. I will wait a few more hours and if they are not visible, I will continue messing with the channels.

Let us know how channel 25 works out. My set-up actually got worse after moving to channel 25. I’m still getting intermittent problems (e.g., my garage and flood lights remained on after sunrise because of the delivery_failed). It doesn’t happen all of the time so at this point it’s just annoying and buggy. Still, I’d like it to be fixed.

Home Assistant development should at least look into an optional switch that when that error comes up for Zigbee-related devices, the connection is continuously checked over-and-over again for a set number of retries until it eventually errors out. This could at least solve the short-term frustration.

1 Like

For now, I had to re-pair 6 door/window opening sensors. They did not migrate automatically to the new channel. The rest of the devices migrated without any problems.
Everything is working now, but I only migrated yesterday, so I’ll test it and let you know next week.

Unfortunately, today the problems with the IKEA sensor started again. Resetting only ZigBee did not help, nor did resetting the entire HA.
I have no idea how to fix this. Unfortunately, my entire wifi band is occupied.

After switching from ZHA channel [auto] to [22], it’s been 48 hours and no failed deliveries yet. Kind of ironic that channel 22 had a higher utilization percentage too.

Back to sucking. My flood lights are all of a sudden unavailable and triggered a Delivery Failed 102 error.

Same problem here. Finally did u manage to solve it?

I’m observing exactly the same behaviour. I own SkyConnect dongle, this is running on supervised HA OS as a VM on ESXI. Normally all works perfectly, just after recent major update all zigbee devices started to disappear, even if they are active ones. Many error 102 delivery messages. Even, I have random light on and off on one of the switches.

I’m out of ideas, as I tried to repeat all what had been suggested in this thread.

So you have any suggestions?

Leszek

I have the same issue.

I have a mix of zigbee brands from IKEA, Ecodim, Sonoff and Tuya knockoff brands. The (end and router) devices are located all over the house. I use a Sonoff ZBBridge over wifi with tasmota and ZHA.

I have changed the channel to channel 22.

energy_scan:

  "11": 99.94584318876281,
  "12": 99.89631182837557,
  "13": 99.92977369505088,
  "14": 99.82571687826758,
  "15": 93.76433891498253,
  "16": 65.26028270288712,
  "17": 82.35373987514762,
  "18": 59.15797905332195,
  "19": 62.257682586134884,
  "20": 59.15797905332195,
  "21": 52.75969252664325,
  "22": 49.512515447068886,
  "23": 68.14622793558128,
  "24": 75.96022321405563,
  "25": 80.38447947821754,
  "26": 73.50699819621309

I have limited the number of end device to the coordinator and have rebuild the zigbee network by turning off the power of the coordinator by an hour.

I have noticed that 102 failure does not occur with the IKEA and ecodim devices. The failure occurs sending a command to the Tuya device’s and my newly installed ZBMiniL2 (end device).

The other way around the state of a device changes directly in HA when for example switching off the smart plug with the button on the device.

Does anyone recognise this behaviour?

Same issue over here. SkyConnect + ZHA, with 2 Apex Smart Sockets and 2 Nedis Climate Sensors.

Everything worked fine for several months, until this issue started a week or two ago. Rebooting HA, the HA VM or the host Win10 doesn’t bring the ZigBee network up. Last time the problem seemed to fix itself with time.
I’m new to the ZigBee, so my troubleshooting skills are currently limited.

Haven’t tried to change the channel, since based on this thread, it doesn’t help.

Does this issue reach the HA devs from this forum, or is a more formal bug report needed?

I have noticed some improvement -for now- :

  • changing the channel
  • switch off the coordinator for half hour and rebuild the network.
  • limit the amount of end devices on the coordinator.

Interesting. I have the same issues with sky connect. I wonder if it IS a chip issue, or a firmware issue. I’ll get a new dongle to move the network into and test. I’ve been trying to keep my sky connect up to date, maybe I should roll back a ways.

Im having the same problem for a while. have to restart HA once a day for my zigbee network to work. My network is already on channel 25. Im using a Sky connect.

Having the same issue there since a few days/weeks.
Im also on Channel 25 using a SkyConnect stick with a USB-Extension Cable on a Intel NUC.

Any ideas? I added a notification to every Automation i control some ZigBee Devices with, if i do not get a notification the automation was failing…

I’m also having this issue. I had a sonoff dongle-e that would just ‘disappear’ (as if not plugged) so i ordered a new one, which stays connected but sometime a device randomly become unreachable and there is a Delivery failed: 102.
My energy scan says :


"energy_scan": {
      "11": 21.09014924761344,
      "12": 59.15797905332195,
      "13": 62.257682586134884,
      "14": 4.69985354430736,
      "15": 55.9836862725909,
      "16": 49.512515447068886,
      "17": 82.35373987514762,
      "18": 75.96022321405563,
      "19": 19.00785284282869,
      "20": 1.5075412082833717,
      "21": 15.32285793082191,
      "22": 70.89933442360993,
      "23": 75.96022321405563,
      "24": 9.713248103580147,
      "25": 7.659755505061292,
      "26": 68.14622793558128
    },

I haven’t changed the default channel, channel 15 doesn’t seems to be busy and this happens in the middle of the night when there are almost no wifi consumption.

The stick is plugged in a powered usb hub, 50 cm away from the raspberry pi 3.

All the switch are on the same fuse, switching them off and on fix the problem for sometime.

I am also on channel 25 and have such problems especially with ikea devices and sometime with my TUYA TRVs.

Is this channel related? So I will switch again but my options are rare since alle channels have 30%-90% utilization.
Wonder if z2m also has such problems.

I used a CC2652P using ZHA for 2 years with no issues whatsoever. It was running whatever firmware was the latest in 2021. The system went through many changes both in software setup and in the hardware wiring and location. I never re-did the home assistant setup because it’s running in a container and I was always able to just move/migrate it.

Now I switched to Sky connect (EFR32MG21) because my old dongle started having electrical issues(I soldered it myself back when I was really bad at it).

And with that I now have a very unreliable connection just like everyone else in this thread. So either it’s a freak coincidence or Sky connect is the issue.

If nobody here proves me wrong I’ll probably buy another CC2652P and report back if the issues are gone again. I won’t get to that until mid January though because I’m on a long vacation now.

I don’t think it is the Skyconnect. I migrated to a TubesZB and I have the same issue. I call a script from node red, which works sometimes and at other times fails. Running the script directly causes the same behavior. I tried just setting the light from node red directly and again it works sometimes and fails sometimes. I have reconfigured my network APs so they reduce the traffic where my zigbee traffic is running. I am at a loss on how to fix this.