I'm seeing odd behavior when I run HA in a container -- something is causing zombie threads to be created. These have the same name as a thread that HA has created, but they shouldn't exist and don't seem to be doing anything.
If you are running HA in a container, can you check to see if you have the same issue? If you get into a shell you can see the threads via "ps -T" or running htop (if installed).
I have seen this on a Mac running HA in Docker -- both Intel and Apple silicon. Have also seen it on an Android device under Termux and udocker. The number of zombie threads created seems to be the number of cores minus 1. That is, an 8 core system will create 7 zombies of a thread that HA created. A four core system -- only 3 zombies.
It doesn't look like HA is doing anything explicit to create these zombie threads. It looks like they might be created by something else in the system/environment.
Thanks! Those three SyncWorker_2 threads (26026, 26027 and 26028) are what I'm calling zombies.
HA creates a pool of SyncWorker threads, but there should each have unique name SW_0, SW_1, SW_2, etc. Something is leading to the creation of zombie/duplicate threads name SyncWorker_2. As far as I can tell, there is no code in HA that would do that. Thus, there must be some system quirk that is doing this counter to any documentation.
It's great to get confirmation of the problem from a different platform!