#ZHA causing repeated Segmentation Faults?

1 messages · Page 1 of 1 (latest)

gray violet Aug 9, 2025, 1:38 PM

For 2 weeks or more I have been struggling with a very instable Home Assistant. It keeps crashing and restarting every so often. I tore down quite a bit of it, deleted stuff, and am starting to feel like a butcher (no offense to the profession). I've exceeded my limits and need help! 😭

This morning I found a segmentation fault in the homeassistant.core debug logs. See next post.

The confusing bit here is that I am quite sure I had disabled ZHA and was still having crashes.

I just looked at ZHA, and all devices were disabled by "CORE config" (not sure), so I disabled the top level (in the cog) and an "Enable" button appeared. I then tried to enable it again and all the devices turned from gray to white but a popup window stated that enabling/disabling the integration failed and HA crashed again with the first sign being "Connection lost. Reconnecting…" and the UI no longer working. After a long wait for it to come back up I used the console in the Proxmox VM to core restart and it came back up very quickly. The logs from this segmentation fault are a couple posts down.

System Details:
Proxmox host: i9-13900 w/ 128GB RAM (3L Lenovo workstation).
HA VM has 20 cores, 16GB RAM, 256GB SSD assigned to it (overkill, I know)
Zigbee stick is a Sonoff Dongle-E (SiLabs chip) with firmware 7.4.4 (I only use it for Zigbee, no thread as I have a dedicated stick for that)
Installation method: Home Assistant OS
Core: 2025.8.0
Supervisor: 2025.08.1
Operating System: 16.1.rc1
Frontend: 20250806.0

📎 segmentation_fault.txt

The Segmentation Fault here is the result of enabling and disabling ZHA (see top post). I had to manually restart core through console (Proxmox VM) as it was not coming back up after quite a long wait.

📎 segmentation_fault_when_enablingdisabling_ZHA.txt

gray violet Aug 11, 2025, 8:46 PM

Ok, taking desperate measures here given there is no answer... I noticed that when I removed an old contact smartthings sensor to replace it with a new zwave one, ZHA started crashing aggressively again, even worse than before. I then started trying "reconfigure" on devices that I had "worked on" recently and ran into an old outlet that crashed ZHA every time I clicked reconfigure. I fed ChatGPT5 zigbee.db and it found an orphaned entry which it said could be a cause of my crashes... maybe the device was the one I just removed or something else from recent changes, anyhow I asked ChatGPT to remove it and it did. I kept a copy of my original file and replaced it with the AI edited one. Once I re-enabled ZHA the network started coming back up fine. I pressed reconfigure on a couple outlets that were causing issues before, and at least for now, it works fine. Time will tell...

And yeah... I know I was brave? or irresponsible? for letting AI modify my db but I kept a backup, and I am getting to the point I will nuke it to start over anyway.

Oh, and the just released ZHA 0.67 and 0.68 in core 8.1 did not seem to make a positive difference but I did not expect one given the changes seemed unrelated.