HA stops responding to events/UI but I see no errors in logs

it seems to work for a half day to almost a day after I hard restart my UTM install. Then after suddenly no sensors or switches update, but the UI is responsive. Light switches clicked in the UI take for 1 second them go back to their previous state and don’t reflect the actual state of switches or sensors. I’ve scanned through logs and can’t find what’s causing this… any clues on where to look or what to try would be appreciated. Ive been using HA for many years. I’m running 2025.5.2 in UTM amd64 on an Intel Mac mini for almost a year now. Was on RPi3 prior.

Are your switch’s or lights polled entities? Are they tuya devices? What integration do they come from?

I have tons of stuff - a boat load of TPlink local polling / integration. A couple of Shelly mini buttons, enphase solar sensors, a Pentair pool integration with lots of switches, appleTVs, a Yamaha receiver… 2 WLED project devices…

The local polling, are you aware that they take upwards of 30 seconds to reflect a state change? Basically anywhere between 0 and the polling frequency.

thanks, yes aware of the local polling delay, this is something much more problematic. My HA becomes unresponsive to all sensors and even some scripts stop working. the Sensors stop updated at some point entirely with many hours lag. HA command line responds in the situation but command line restart or reboot of HA is met with ‘not ready’, cannot reboot from UI, I have to restart my vm to get everything to work and it seems solid after a reboot for many hours. I need to find some logs that explain what’s corrupt or broken? not sure what integration or code is causing this - it used to be super stable for me. so frustrating.

when I type reboot or shutdown, I get the response from the terminal :slight_smile: “System is not ready with state: shutdown”
huh ?

Figured it out !! In case anyone else is running UTM / QEMU on MacOSX, it appears that Hassio running in this emulation environment is not able to sync the time clock and over a few hours the clock skews far enough that events and triggers no longer work. In fact, just ssh into the Hassio UTM and type date several times reveals the local date+time is not even incrementing in any normal linear fashion and sometimes even goes backwards. Restarting UTM and Hassio works great for a few hours, then stops processing events and becomes unresponsive slowly. Guess I’ll try running a container on Linux instead, or maybe go back to my RPi config I ran before.