Zigbee super unreliable?

One other thing which is sometimes forgot. And, it might not be relevant for you. The mains (powered devices), they should have power all the time. Not being switched off and on to control the light. The zigbee network can be really confused… even it should not😜

To be honest I think the WiFi interference topic is a distraction. Devices that communicate ok then drop off after a few months are nowhere near anything else using WiFi.

I will check what channel WiFi I’m using.

What I don’t understand is why devices completely drop off the network. I get that there could be interference and some lost messages, but why does a device give up and stop trying?

It’s like devices forget about each other and need re-pairing to remind each other that they exist. That seems a protocol/sw issue to me.

I would really like more diagnostic visibility, e.g ‘the signal from x device is so weak that we have taken it off the network.’
Or
‘We have not heard from x device for y days so we think it has stopped working.’

Instead, just ‘unavailable’ and a dashboard with lots of error messages.

I started out 2 years ago on a Conbee II directly attached to an RPi4. Then I learned to use a USB extender, and I have even got longer. There is more than 2m from anything electrical to my coordinator, well hidden in a plant.
This provided a reliable zigbee network, however still with some strange stuff happening, like changing a lot of devices. Got a new coordinator (sonoff 3.0 plus) and this last part was also solved.

My zigbee network have been super reliable for a full year, except when I during the summer tried to install an updated beta firmware. Downgraded again and everything was fixed. Or, when I got 2 super cheap bulbs from AWOS, however they reported every 1 second, causing problems. Turned them off after 2 days, and everything was fine again.

I have along the way decided on a few rules, which work well for me… This is in no way based on anything, except my own belief. I do not use Sonoff devices (except the coordinator) as the first few battery powered devices was dropping the network (i know this was 2 years ago, however I have not changed my mind:-). I never use Tuya battery devices, they are hard to get on the network and behave strange. I have Tuya plugs, working well.
I have a lot of aqara devices. The battery devices I always connect to an aqara plug or aqara relay, otherwise they have a tendency to drop at some point and needing re-pair. They also need re-pair if I move them or the plugs, as they are not smart enough to roam. I have decided to live with this, as they are cheap:-).
I have been through a few super cheap devices from China, without names. Some worked and some did not, however I decided I would not waste time testing it, hence stopped buying them.

So, a long story. The conclusion, a zigbee network can be reliable, to the extend I do never start zigbee2mqtt webpage just to check. I only open it when I need to make a change.

There is a bit more if you use deconz you can see a network map of what’s connected to where, but its still a bit nonsensical and PITA to debug. The only perk is I can then often track down which mains powered device needs rebooting faster that way.

I think I’m noticing it might be more of an issue when we have storms or blips in power that they are more likely to have some devices stop working. I still don’t understand why though – in one case a mains powered repeater was on a UPS and still quit working after a power hit (and yes, the UPS works - nothing else on it was affected this way, only the Zigbee device). Furthermore, it seems like the problem is worse if the power just “flickers” or “blinks” vs if it stays out for say 15+ seconds (which is far less common in my area). I have also found that a “quick” way to fix everything is to just throw the house main breaker off for 15+ seconds and back on again to reboot all the Zigbee mains powered devices at once (and removed the one from a UPS) seems to cure the outage 100% of the time so far. It can’t be the gateway because my USB stick is on a server with a UPS and never restarts when I cycle the power to everything else.

The thing I don’t understand is why these supposedly smart-home things can’t cope with normal things that happen in a home electrical system with power anomalies, and apparently can’t self-heal afterwards. All my Z-Wave switches and things resume working again after their next commanded action if they drop at all.