I’m trying to add sign-in with google to my application. I got it working locally based on official documentation, but I’m struggling with enabling it on real host on internet. I put this domain address in API Services → Credentials → API Keys → Authorised JavaScript origins
When popup window appear Javascript generates request: GET
I will answer myself: It only works for https. So after adding certificate and puting web traffic through it. I change addresses to https and it works. Response message could be more descriptive about that
I’ve experienced this issue on a web app that has been working flawlessly for a number of years. Since a few days ago, a few users have been unable to access the app with the “Sign in with Google” option, but now the problem seems to be more widespread and is affecting nearly all users. The login URI is all over HTTPS, certificates are all valid and no changes to either the login code or the app have been made in the last month.
I am also seeing this same issue. Apps across all of our environments are now failing Google OAuth login as of the last few days. Nothing in our code has changed and I verified all keys are correct.
EDIT: Interestingly, our local environment using http://localhost:4000 appears to be working, but all dev and production apps have stopped working.
Same. Did you get an email about the FedCM migration? We did and it stated that: (1) it would happen in April and (2) no issues were detected for our site/project. Apparently neither is true. I followed the migration steps and still not working.
The response uri did not have the full path to the endpoint that received
the response - it only had the host. On changing it, my issue was resolved
in minutes.
I was also able to temporarily fix the issue by updating the Authorized redirect URIs on the Google Credentials screen to be the exact path to the redirect URI. In our case this is https://example.com/api/v1/auth/google/{clientId}
While this solution works for 1 client, the URIs do not support wildcards and the redirect URI will have a different path for each client…
Was this an intentional change, or just a bug? Seems weird they would make such a massive change without informing users and breaking a ton of sites.
EDIT: I got it working by passing the client ID on the button’s state property instead of the redirect URL, which allowed me to redirect to the correct client domain.
Ours is working now. The redirect_uri had to be updated to include the full path, e.g., [https://myhost.com/login](https://myhost.com/login). Before this, the trailing /login wasn’t required. Hope this helps.
