Possibly caused by Gitlab changes in their latest release, e.g API changes, that typically happens at the end of each month, the 22, although that seemed to work fine 6 days ago (24).
Probably not the cause of the problem but because of docker images, the codecov uploader version I use is not necessarily up-to-date. In the present case it is the version that was live on the 11 of August.
@TungstenHeart it looks like this might be a lingering issue from this topic. The issue I’m seeing is that there is not refresh token information from the bot, and so we are unable to retrieve the necessary information from GitLab.
This is still broken. So one monht ago, what you suggested to try had not work.
Today i’ve tried
delete data ( )
revoke codecov, accept it again
other user name as team bot
use latest uploader, not the one from the docker image
and it is still broken. @tom I have two questions:
Cant you fix that manually ?
Are the API requests logged (thos sent by your service) and could you check that the addresses are good ? Actually upload is correct, analysis of the gcov files is too. The problem seems to be simply that requests are not done at the right place.
I’ve noticed something. Maybe the problem is that the your service expects the team bot to have the “admin” role. But this cant work on gitlab.com, only on the self-hosted gitlab. On gitlab.com users own their repos, have write access, etc. but are never “admin”.
I see myself as a “developper” here Codecov, this is also the team bot account for the org… which cant be admin.