Thanks for finding that reference
I ended up reverting HA to 2025.2.4 from a backup. I’m still seeing the duplicate devices, but it looks like there’s a set of steps to resolve that in that Github discussion which I didn’t get a chance to yet
Additionally, only one of my plugs stopped collecting data - and it was the first one I ever bought. HA was receiving the current_power metric from it after the 2025.3.x update, but the switch on/off toggle was not making it into HA. This makes complete sense, since that one plug has some weird version of firmware different from the others, so I’m assuming that the pyvesync lib refactors inadvertently broke the switch power toggle for that switch only. I will try to fix the duplicate issue and then see if I can work with maintainers of pyvesync to get that oddballl firmware plug working, and then wait for it to make it into HA before upgrading again.
I also work in software. These kinds of breakages happen all. the. time.