Mattia Verga

Changing to "high trouble" because, when I tried to start working on that, I faced a lot of troubles I didn't thought before: what if the update is a multi-build...

While trying to update bodhi docs regarding update states, I realized one possible flaw. Updates in pending moving to testing are not gated, unless they're critpath updates. For "normal" updates...

Can we consider this fixed by #5538 ?

Maybe it might be better to unpush an update that reaches the negative karma, rather than obsolete it? But then we have to be sure it cannot be pushed again...

Another case is https://bodhi.fedoraproject.org/updates/FEDORA-EPEL-2024-c154b725ab

Bodhi has some sort of modular approach to interface to bug trackers, most of the code is in `bugs.py` in the `Bugzilla` class, so I think it can be extended...

Not sure what's going on there. The openid authentication seems to be handled totally by pyramid-fas-openid and I cannot even see any line in the logs when the HTTP 500...

Yeah, I see it was handled in https://github.com/fedora-infra/bodhi/issues/4356 but pyramid_fas_openid is still used for `login?method=openid` calls... should we drop openid authentication?

From further investigation, this is indeed what is causing server hiccups, at least before we added a query timeout setting. The problem here (which I can reproduce in the local...

As I read the unit test, it is: https://github.com/fedora-infra/bodhi/blob/f610256f15b245e530664fb0e985f76be467fdb4/bodhi-server/tests/services/test_updates.py#L1546-L1558 creates three users: `bob` is a provenpackager, `ralph` is a normal packager, `someuser` is neither of both. (yes, the use of...