Nabu-Casa Remote UI not working for Google Integration

So I’ve been using Nabu Casa Cloud flawlessly to integrate HA into Google for voice commands for 9mo since I moved to 5G internet and no longer had a static IP. However, a few days ago its failed and I’ve not been able to restore it.

Its very weird, the remote UI is clearly working as I can access my instance from the Alternative name address, however google when I try to reconnect ether respond with an error has occurred or that it could not reach HA Cloud.
image

Interestingly when I log into Nabu Casa it also does not believe its connected to my HA instance:
image

However from HA it looks like its connected:

That being said, my log shows an error about not being able to retrieve my sub status:
image

However, that was from the full system reboot I performed upon request by Nabu-Casa support.

So I’ve tried all the usual suspects:

  • HA reboots
  • System reboot
  • Logging out and back in to Nabu Casa Cloud
  • Relinking from Google Assistant

I’ve not made any changes to my config, however I did update to 2023.11.3 earlier in the week and I also noted that my remote UI certificate has been renewed a few days ago too.

I already have an open Ticket with Nabu-Casa in case its on their end, however they just about take the full 24 hours to reply with a “reboot your system” and let see if that fixes anything.

Any ideas where to next to debug further?

Well I’ll post the resolution here.

The issue was a aiohttp flood error being caused by an integration bridge phone that was connected using HA cloud (https) instead of the local address (http) and the dns rewrite was pushing the https request to the http.

How or Why this caused the issues is beyond my knowledge but its fixed now. Also why its suddenly an issue despite no changes being made is also a loss to me.

Hopefully this will help someone else along the way, i’ve spent over a day debugging, even restore an older back up to find it was still an issue. Along with breaking multiple integrations and having to fix them too (the worst of is frigate that decided not to wait for the mounted external folder to be ready and created a local one and filled the vm hard drive… doh).