Problem solved. “ Manual setup (if you don’t have Home Assistant Cloud)” Google Assistant without Nabu Cloud works.
My problem was Duckdns related to my internet provider.
The ip4 address is just a nat address. This causes Google to try to access my HA via the ip4 address. I could try my luck, but then there was no response from Google.
The solution was incredibly simple. The ip4 on Duckdns was deleted and an automatic update was prevented.
It’s been running for 2 days without any problems. But now I have 2 new ones  My VPN WireGuard® is causing problems. The IP4 is gone and will probably have to display it again for the Fritzbox. And on Google I get:
Add capabilities
Seamless Setup Support local query
not selected.
But I’ll find that too. It doesn’t work, it doesn’t exist.
diagnosis update 2 : Solution
issue with ip and dns resolution
yes duck dns , no ip like other services can help you to get access over internet but if you using ipv6 only then your google home app show error "could not reach [test] all
solution you need ipv4 as well to resolve ipv4 if you don’t have ipv4 like me here is solution for that as well
buy domain or if you have domain configure with cloud flare it will resolve your ipv4 queries and transfer to you on you over tunnel
before you setup cloud flare make sure you turned of reverse proxy like nginx and duck dns or no ip or what ever you are using
setup Cloudflare in home assistant
then google home and Alexa will work without any error
I am currently trying to set up HA with Google HA. I’ve got a domain and my HA instance is reachable from outside. I’ve configured the Google part and configuration.yaml like in the official docs but I am not able to add the “Device” in the Google Home app. There i get an error 404 on …/api/auth/authorize. Out of curiosity i checked /api/auth/token (wich also get a 404) and /api/google_assistant (which replys with "405: Method Not Allowed).
Now i found this thread and wanted to to add the Directory information in the google action. And the next problem was born… I am not able to upload a Logo. I’ve tried different image formats with diferent resolutions and different file sizes on differnt devices and on different networks.
This google assistant setup is driving me crazy, am i the only one who get this error?
I’m also getting the “Could not reach…” error when trying to re-connect to my HA instance. My GA integration stopped working in the last day or so, so I unlinked the account and am now unable to re-link it. I’m using the manual GA setup, not nabu casa.
I tried the idea of exposing a single, simple entity, as suggested above, but it didn’t help.
???
Adding my voice to the crowd - getting the same thing. Just tried to set up following the “manual” instructions. When I try to add the [test]… app from the Google Home app, it takes me to my home assistant login page using the external URL, I am able to log in, then I get a white screen with a (Google-themed) spinning wheel for a few seconds before it drops me back to the list of apps with “Could not connect to [test]… Please try again later”.
Neither the Google Cloud logs nor the Home Assistant logs show anything. What are some additional troubleshooting steps to try?
I’ve got the same issue here, but in in interest of completeness in a long thread:
Using nginix reverse proxy, with cloudflare dns proxying
Have has this setup for > 2 years working, “some time ago” I noticed new devices were not being added during a sync.
Local fulfillment turned on
Unlinked and attempt relink
When I attempt to link, get the usual login screen, 2fa check then “Linking your ” followed by the could not reach error.
In the google log explorer I get whenever I attempt to link.
SYNC: Request ID update devices failed: CANCELLED
I’ve tried disabling cloudflare proxying, setting it to expose one entity, and setting the project back up from scratch, strongly implying its something to do with some part of my setup - same result from both.
My guess is that there is something network related going on but that doesn’t disrupt standard “out of home” access, maybe combined with a change on the setup side.
For me I’m guessing either cloudflare or NGINIX or some combination of both. I’m interested in the other people with this error, are you guys all using reverse proxies?
It seems this thread is awash with things like duckdns issues etc, and there are a few commonalities.
My intention when I have some free time is to try and switch things around - probably trying cloudflare tunnels and no nginix at least as a debug step.
I don’t think it will resolve it no - as I understand there has been some problems at duckdns over the last year or so that have led to some instability. I know it ‘can work’ using that method but the issues they have had has been disrupting connections from outside the home in general, including from Google. That has been ‘some peoples’ issues in this thread, but not all of them are in that category.
I’ll report back when I’ve done some debugging and maybe I can figure out at least which bit isn’t working.
I did try linking it again today to see if I was in the “it just started working after 24h” category but this isn’t the case. I’ll try more on this (I expect) next week.
Much to my surprise, I got back from a trip this weekend and decided to relink again with no other changes and it just worked. This makes no sense to me, and sorry as this won’t help others with this issue. The only thing I’ve actually changed is to recreate the project from scratch. What I haven’t done:
Changed the logo / developer information (tried that on the first project, but did no the second)
Change any proxy config from nginx or fiddle with anything at cloudflare
I haven’t had any changes to internet connecivity, DNS or network.
I’ve no idea why my initial project ‘failed’ after a period (maybe it is something to do with it not being a fully published one but that feels a little unsatisfying).
Either way, all I did was create a new one and wait 3 or 4 days. Either it intermittently works 5% of the time or its a waiting game on the new project I guess.
Is the code owner aware of this problem? Seems to me that there are different flavours of problem but probably something common in the cause. There must be someone who can definitively advise which logs to look at for which messages.
For me, I’ve had problems for ages with voice commands to HA over Google Assistant intermittently not being available. I thought I’d bite the bullet and build a new home assistant and new Google project to try to define the problem and I now have the inability to link the account “Could not reach [My HA server], please try later”.
I’ve been through the Google Project settings and everything is correct as far as I can see. I’ve tried some of the fixes from this thread but I still can’t get connected. It must be something in the Google account as I can successfully link the old project. So here’s goes another hour or so re-checking the settings…
EDIT: This is weird as I can findthe new project in my list of “Works with Google” stuff.
I get the username/password screen which presumably is local.
If I put the correct uid/pw in I get “could not reach” after it tries to connect.
If I put a bad p/w in it tells me it’s bad, so it MUST be reaching.
Very strange, this did the trick for me too for now. Experience is that every couple of months there is a new issue with this integration, but for now it seems to work!
Thanks!
I have the exact same error logs 1 year later. I tried unlinking project, deleting it from Google Cloud starting fresh. Same error. Did you ever find a fix?
I was also dealing with the “Could not reach…” error. I can confirm the issue was Cloudflare Bot Fight Mode.
I didn’t disable the Bot Fight Mode, but I added AS15169 as allowed in the Cloudflare Security WAF.
I had a working google assistant over nginx; however since I took out Nabu Casa I decided to delete it. Now the problems occur…
My log continues to show:
“Error setting up entry home-4481a for google_assistant”
Peculiar thing is that I do not have a config entry anymore, I deleted the JSON, all references in the config files etc. Anyone got a clue what is happening here?
