Really can't figure this out. It ought to have worked. I can't reproduce it right now; if I bodge up the script to just show me what it's doing, it shows me that it would create new bugs now as release blocking if I let it.

I'm gonna add better logging to this stuff so we can catch it next time, and leave it for now.

Better logging.

OK, so I spent more time on this today and finally figured it out. It seems to be a permissions problem. The Bugzilla account that is being used to create these bugs in production does not seem to have the necessary permissions to mark a bug as blocking another bug.

I wrote a test script that creates a bug and tries to mark it as blocking the F36 Beta and Final blocker trackers. If I run that script in a context where python-bugzilla is authenticated to my own account, it works as expected - it creates a bug that blocks those trackers. But if I run the exact same script on the box that runs relval size-check in production, using the "Fedora QA Tools SIG" Bugzilla account, it creates a bug that doesn't block anything.

So, I think we need to get that account's privileges elevated to make this work. I'll ask infra to do that.

This should now (finally) be fixed, as shown by https://bugzilla.redhat.com/show_bug.cgi?id=2041648 .