ZHA - help needed - stable network got somehow unstable/unusable

[marc4s]

The problem

Since 3 years I have HA with zha(with sonoff donlgep channel 15). it was almost (except some small bugs in some releases long time ago) really stable.
But somehow it got really unstable /unusable since end of 2024/beginning2025

there was no new HA release which caused this as it started by around end of december, so there was around 3 stable weeks with 2024.12

the behaviour:

1. ALL zha devices getting "unavailble" / zha is reinitializing which takes a long time
2. after initialising is done all seems to be visible/available again - but changing the state of a switch/light is not working most of the time. "no response from device" sometimes after this zha is immeditaly reinitialising itself (but not all of the time). somtimes after the timeout the changed device is changing its state (for example a light is getting on)
3. sometimes (I am still not sure why/when) somehow the network is getting stable, fast,responsible. and then over night without interacting it is completly lost
   see in this screenshot - at the left stable for a few hours. then a lot of self reinitialisings. then over night completly lost

in the last 2/3 months I already tried:

• switching some devices to Z2M (conbee III with zigbee-channel 11) to get them out of the zha network
• wifi is on wifi-channel 11 (so far enough from zigbee-channel 15)
• tried other z-stack firmware (started with 20230507, no success with 20240710 and currently on 20221226) without success
• deleted/removed devices from network when I believed they are causing this behaviour when I touched them(changed their states) - also without success. then other devices did the same

Somehow I got lost
i am desperate and would be very grateful for further tips or maybe someone can see a reason in the debuglog

home-assistant_zha_2025-03-06T07-43-24.805Z.log.tar.gz

I did not completly switched from zha to z2m as I was so happy with zha for a long time and I could not successful pair some devices with z2m (lidl bulbs, even if they are offically supported). so maybe these bulbs are the faulty devices. but I need to know this clearly if it is like that or if it is some other device.

ah and I think 2 or 3 weeks ago I had an issue with the zigbee.db file. it looked like the database got corrupted and zha is not starting anymore. but the error was gone after (maybe ~20-30 retries)
so if the zigbee.db is corrupted maybe it could be fixed?

What version of Home Assistant Core has the issue?

core-2025.3.0

What was the last working version of Home Assistant Core?

core-2024.11.4

What type of installation are you running?

Home Assistant OS

Integration causing the issue

zha

Link to integration documentation on our website

No response

Diagnostics information

No response

Example YAML snippet

Anything in the logs that might be useful for us?

Additional information

No response

[mib1185]

Hi @marc4s

We use GitHub for tracking issues with Home Assistant Core itself, not for providing support - for support request you should try our Community Forum or join our Discord chat server.

Thanks! 👍

[marc4s]

for me it is not clear if there is not a bug. what should it else be if it was working for a long time

[mib1185]

there was no new HA release which caused this as it started by around end of december, so there was around 3 stable weeks with 2024.12

at least this is an indicator, that it is not an HA core related issue, so as a starting point asking for support on the proper support channels, would be the best option.

[marc4s]

how much time may elapse before a certain state occurs that calls up a faulty code/leads to incorrect behaviour?
it's new to me that a certain amount of time has to pass to say, hey, that's no longer an error. because then there was never a y2k problem and there will never be a year 2038 problem. until then it was running

of course i'm asking for help here and not in a coffee party forum. because i want to fix the error professionally. sorry for the comparison.

[dawid-woitaschek]

@marc4s I do have the ConBee III as well and with the installation of the first HA/ZHA updates in 2025 the whole zha stuff is unreliable as hell.

Yes, my zha is reinitializing as well at random times, but doesn't recover by itself. Sometimes the zha integration can't communicate with the ConBee, for example after selecting a data flow mechanism (?) like hardware/software/the "emtpy" selection.

This freaks me out... Especially after HA updates it never recovers and I have to reboot Home Assistant or replug the stick.

Dunno if the devs around zha messed something up or "dresden elektronik" has to deliver a firmware update. But it's weird that it worked like a charm for months and now it doesn't.

I enabled the debug log for zha and downloaded it after it became unresponsive but still searching through as this is around 45 mb.

[marc4s]

I have ~80 zigbee devices. mostly they act as router (main-powered bulbs and switches). and as the network was stable a long time I was happy.
currently I "resolved" the behaviour by splitting the network into two as written above. zha got stable as I moved ~15 devices to z2m.
so currently zha is stable and fast and I would not recommend anyone having more then 60devices in there.

knock on wood - it is stable since 1 week

so both networks look like this

zha/sonoff stick - 55 devices
z2m/conbeeIII - 20 devices

I have some remaining devices which I may connect step by step in the next weeks and watching if it is still stable

[dawid-woitaschek]

@marc4s Do you have DeConz installed as well? Came across #136077, where this was the culprit.

I did - removed it now and will observe this again.

[issue-triage-workflows]

There hasn't been any activity on this issue recently. Due to the high number of incoming GitHub notifications, we have to clean some of the old issues, as many of them have already been resolved with the latest updates.
Please make sure to update to the latest Home Assistant version and check if that solves the issue. Let us know if that works for you by adding a comment 👍
This issue has now been marked as stale and will be closed if no further activity occurs. Thank you for your contributions.