The /users/#{username}.json endpoint intermittently fails with “URL not available” (using the API gem), and
The user’s smaller profile picture is missing:
Otherwise, we cannot see anything wrong about this user. In fact, we recently set up a staging server populating data from a backup of our live system, and the user is fine on the staging server.
Any way to do a data integrity check to see if there is a data issue?
We’ve seen several reports of this related to SSO. Does your site use SSO? It looks like something in the way the SSO call is constructed is changing the user’s data at login time.
Very interesting. Turns out we’re developing a membership system on top of Procourse Memberships. The staging server is staging that development work w/SSO turn OFF and the live system (where we’re getting the error) still has SSO turned ON. The good news (for us) is that SSO will be OFF on our live systems in just a few days when we go live. We can live the issue (effecting just one user) until then.
For others, here’s the error as received by the api gem: DiscourseApi::NotFoundError, {"errors"=>["The requested URL or resource could not be found."], "error_type"=>"not_found"}. As suspect as it sounds, it only happens like every other time the call is attempted, and then only on the same single one in 250 users.
Any suggestions in terms of fixing a corrupt user, export / import, copy, or does one need to delete and start over? I assume there is some recommended path to correcting a problem record?
SOLVED: Seem to have fixed this user by retyping a new user name, yet then going back to the old username then seemed to work fine thereafter. I say “retyping” because earlier attempts to make a small change to the end of the user name failed. Not sure, but suspect perhaps a strange, invisible character got into the username? Also, this user had (and still has) an extra cap other usernames don’t have, e.g. First_McLastname.