Telldus Live (local API) breaks after Reload or Disable/Enable – "Config entry has already been setup"

Hi,

I believe I have found a reproducible issue with the built-in Telldus Live integration when using the local API.

Environment

  • Home Assistant Core 2026.7.4
  • Home Assistant Container (Docker) running on a Synology NAS
  • TellStick ZNet
  • Local API connection configured with:

tellduslive:
host: 192.168.x.x

No Telldus cloud polling is used.

The problem
After a normal Home Assistant restart everything works correctly.

However, if I either:

  • click “Reload” on the Telldus integration, or
  • Disable the integration and then Enable it again,
    the integration immediately stops working.

All Telldus sensors stop updating until I restart Home Assistant.

A full Home Assistant restart always restores normal operation.

The first error is always:
ValueError: Config entry (…) for tellduslive.sensor has already been setup!

Depending on how the failure occurs, I may also get many duplicate entity messages such as:
Platform tellduslive does not generate unique IDs. ID _222-temp-0 already exists

Additional observation
Before I discovered that Reload reproduced the issue, I experienced exactly the same failure once during normal operation after approximately 12 hours of uptime.

I had not manually reloaded, disabled or modified the integration.

The symptoms were identical:

  • all Telldus sensors stopped updating,
  • the same “already been setup” error appeared,
  • restarting Home Assistant restored normal operation.

This makes me wonder whether Home Assistant or the integration sometimes performs an internal reload or reinitialization that ends up in the same faulty state.

What I have already checked

  • Only one Telldus integration is configured.
  • No duplicate configuration entries exist.
  • The issue is independent of entity names.
  • The issue is fully reproducible using Reload.
  • The issue is also fully reproducible using Disable → Enable.
  • A complete Home Assistant restart always restores operation.

Has anyone else seen this behaviour?

If needed, I can provide the complete traceback, debug logs and perform additional tests.

(the debugging and text was made with the help of chat gpt, and manually verified)

Issues are all entered in GitHub, not here. This will not be seen.
¯\(ツ)/¯ Sounds like you are looking for…