ERROR invalid_grant

[davidwagner]

We have dozens of students in Data 8 who have experienced

ERROR | auth.py:102 | {'error': 'invalid_grant'}

when trying to log into okpy from a Jupyter notebook (_ = ok.auth(inline=True)). I can't track down any commonality into what is causing this. The workaround/hack we have been using is to add force=True, but that is not ideal because then it requires them to reauthenticate every time. Also it seems like a kludge to paper over an issue, and it would be nice to understand what is causing this problem and fix it.

How would I go about debugging / troubleshooting this?

[rahularya50]

I'd suggest following the instructions in the README to get a copy of OKPy running locally that's built from this repo. Then put some print statements in https://github.com/okpy/ok-client/blob/master/client/utils/auth.py to see where reauthentication is failing.

[rahularya50]

It might be that the reauthentication process does not take into consideration Jupyter notebooks, since that error is what you get when your token has expired on the CLI client.

[rahularya50]

        try:
            access_token = refresh_local_token(server)
        except OAuthException as e:
            # Account for Invalid Grant Error During make_token_post
            if not silent:
                raise e
            return notebook_authenticate(cmd_args, force=True, silent=False)

is what might not be working. Check to see if it's even reaching this point?