DataUpdateCoordinator-based integrations become unavailable after a few hours

Hi everyone,

I’m experiencing the same issue reported in GitHub (#157017), and it’s still happening in Home Assistant 2026.2.1 after migrating from 2025.10.

Several integrations using DataUpdateCoordinator (Met.no, OpenWeatherMap, Open-Meteo) become unavailable after a few hours or a day and do not recover automatically. Restarting Home Assistant fixes them temporarily, but the problem returns.

I’m running Home Assistant Core in a Docker container on a mini PC with Ubuntu Server using Docker Compose. Internet connectivity works fine, and I can reach the APIs from the container using wget, but the integrations stop updating. Reloading the integration manually often hangs or fails.

It seems related to DataUpdateCoordinator and how it handles sessions or caching.

Is anyone else still experiencing this after migrating from 2025.10? Are there any temporary workarounds while waiting for an official fix?

Thanks for any help or suggestions!