Entities become unavailable

[sjassio]

My KNX integration with HomeAssistant has been rock solid for 18 months, but lately I saw these notifications in the history of all KNX entities in Home Assistant. You can notice that an entity becomes unavailable and quickly restores.

Looking at my logs, I see this warning

Logger: xknx.log
Source: /usr/local/lib/python3.13/site-packages/xknx/io/tunnel.py:361
First occurred: 29 December 2025 at 23:30:56 (13 occurrences)
Last logged: 12:29:47
Received DisconnectRequest from tunnelling server.

I have been troubleshooting this by resetting my IP interface. This is a KNX IP Interface 731 from Weinzierl. It's on Tunneling and UDP - rate limit 30 and message history of 200 messages.

I have been playing with my connection settings but this device doesn't allow another setup than UDP.

My network is Unifi based, recent setup. The devices are for security reasons in 2 VLANS, with the restriction that only HomeAsisstant can talk to the VLAN of the IP gateway.

In the mean time to trouble shoot further:

• I moved devices in the same VLAN
• I changed gateways (other physical gateway)
• I played with the settings for state updates

This is also being discussed on Reddit here.

Does anyone have experience in troubleshooting this further?

[farmio]

Hi 👋!

Have a look here: home-assistant/core#153892

Happy new year!

[sjassio]

thanks, I informed them about this thread.

[sjassio]

OK time for an update. I read on some other places to test home assistant safe mode to see if the issue persists. I tried it and it's rock solid when I'm running in safe mode. This means that an integration or automation is probably the cause and not Home Assistant update xxxxx.

Not sure how to investigate further yet, any suggestions to find the culprit?

[farmio]

Safe mode loads Home Assistant Core, but no custom integrations, no custom cards, and no custom themes.

I'd go back to normal mode again and disable half of your custom components. If the issue comes back, try disabling the other half...
Maybe also dig through your logs to see if you can find some hint about which integration could be problematic.

[sjassio]

Just a quick update - running in Home Assistant safe mode resolved the issue, so I rebooted to normal mode and disabled 9 integrations. Since disabling these, I had the same result as in safe mode. I tried enabling 2, but I got hiccups again; however, they were very limited (once every 6 hours). So my next research is whether it's really one of these integrations or if something else is causing a hiccup in Home Assistant.

I'm using a Home Assistant Yellow - could performance be a root cause here? E.g. maxed out memory/database/processor/... rather than 1 specific integration that is acting up?

[farmio]

I don't think so, but you never know... do you have performance issues or maxed out memory or cpu?

[sjassio]

I have been monitoring mem and proc, but both are fine.

• running in safe mode works perfectly, not a single hiccup
• running in normal mode with 9 (well known integrations) disabled, works fine as well

This proves to me that it's not a network issue.

As soon as I start enabling integrations again - it doesn't really matter which one - hiccups start appearing within the hour.

I have no clue where to search next. Any suggestions?

[sjassio]

In the mean time I restored a backup to other hardware (virtual machine) - amount of disconnects is less but after slightly over 4h I had my first disconnect again