Having the exact same issue. This was working previously for over a year on 5 different casted Google Hub’s around my house but now the streams are freezing entirely, do not appear or are many minutes/hours behind.
I’m going to roll back for now until this is fixed
The areas can be recorded in a separate table. Circles need to be converted into consecutive segments to do spatial indexing on them as well. Once a spatial index exists, the database engine can do the job of efficiently finding the areas that contain the position.
I thing my issue with rtsp streams has to do something with updated HLS.js package used in frontend.
Issue comes up when updated to 2024.2 but only when new frontend is loaded by notification that there is a new frontend to load… Then page force refreshed by home assistant, and then issue is popping up.
Thanks Tom, I am aware of the 255 character limitation, I am using the standard HA modbus integration and I just mentioned the 3rd party template, but that is just a template. Isn’t the modbus integration itself part of HA core? I am copying my config below. It only reads “count: 38” and nothing changed between 2024.1 and 2024.2 in that configuration, so I am wondering what actually changed? Thanks.
Okay, did some more digging and it looks like “count: 38” and the python struct: ">2h15H22B10H" actually leads to 285 Bytes and 2024.1 apparently truncated it or did something else and therefore it wasn’t a problem. 2024.2 now throws an error. I changed it to “count: 31” and structure: ">2h15H22B3H" which still gives me all the template sensor values out of the payload that I need and that now is less than 255 bytes and works fine. Sorry about the noise, but maybe it actually helps someone else as well who runs into similar issues. Thanks!
Does the MQTT breaking change refer to entries in the config yaml file or something else? I only have two mgtt_room entries for espresense in my config so I think I’m OK, but just want to be sure.
Still, i can get devices by only one account, SmartLife or Tuya, not both.
The one that works is the one that i reload last.
Can’t get both to work at same time
it remains the original ‘on’ version, while we set it to change in the domain icons before.
I somehow missed that in the beta. (most likely because I have most of my interface with custom:button-card and dedicated icons there.)
this was ok before though.
update
I now see they lost/dont have the device_class: switch, even though the affected devices are core HomeKit or Shelly switches. how odd, I seem to recall they all had that from the moment of being integrated in the their integrations.
the HomeKit seems to be correct in not using device_class yet. (I moved those from Ikea, where they had it set)
Shelly still makes me wonder if that was the case before.
filed an issue in case the icon migration in 2024.2 just failed on this one.
checkin this a bit more, I see all of the integrations in my config killed the device_class: switch, except Hue and Unifi…?
Same here, I can put them in my language on the phone but they are translated into german on HA. Meaning in order to modify the list from HA I need to know the german names of things
Who can we ping to check this out?
don’t know if it’s related to the update but since yesterday, my HA has been “hanging”.
I can no longer connect to it via the web (wan & lan). The container is not stopped, I see the logs continue to scroll but no access.
forced to restart, via portainer.