I get some strange behaviour while trying to log in on the hepdata.net portal. Logging in through CERN fails with some Oauth2 error (even after I manually delete all cookies), then trying to recover my account password is successful in resetting the password (never had one!) and using these credential I end up on the default portal error page - however I notice I am logged in, but I’m a bite reluctant to do anything since this is not the authentication method I used previously and I really do not want to mess things up by making changes with different credentials.
Could you please, investigate and let me/us know what is the safe way to proceed?
Thanks in advance.
Thanks for the message, Alex. It looks like there was a general outage of the CERN Legacy OAuth service (used by hepdata.net) yesterday. It should be working again now. We plan to upgrade to the new CERN SSO in the next week or so.
Regarding the problem of landing on an error page when logging in with email address and password, this is a known issue for an account that was previously connected to CERN OAuth. The cause seems to be a bug in the invenio-oauthclient package used by the HEPData software. I’ll contact the
invenio-oauthclient authors to see if they can fix this bug soon. In the meantime, after logging in, you can navigate from the error page to the Dashboard or any other HEPData page and you will still be logged in without any bad consequences. Alternatively, you shouldn’t see the error page if you go back to logging in via CERN OAuth instead of email address and password. It is the same HEPData account whichever method you use to log in.
I raised this issue with the invenio-oauthclient developers and they released a new v1.5.1 last week with the bug fixed. Today we’ve deployed the new version of
invenio-oauthclient on hepdata.net and we’ve also upgraded to the new CERN SSO.
Hi Graeme, Thanks for keeping me in the loop. I just tested the new oauth mechanism to log in through CERN SSO and it works fine (even quicker than the older SSO in my opinion). I think this issue is now closed. Thanks again!