I think I figured out why bodhi is trying to mash f13-updates (maybe).
There seems to be a bug in the repo detection when we resume. The
code looks like:
for update in self.updates:
release = update.release
# [No Frozen Rawhide] Don't mash stable repos for pending releases
if update.request == 'stable' and release.locked:
continue
if self.resume:
self.repos.add(release.stable_repo)
self.repos.add(release.testing_repo)
elif update.request == 'stable':
self.repos.add(release.stable_repo)
<snip>
So I think what happened is that on the first push attempt it
correctly skipped f13-updates. However, since the bodhi configs
were all horked due to permissions, we had to resume the push.
The above code seems to skip f13-updates initially, but then
blindly add it in the resume case (I think).
I'm not sure we actually want to mash f13-updates right now. I
could create the f13-updates.mash file really easily, but that
seems like the incorrect action to do.
If bodhi is fixed for this bug, we could roll out a new one and
then resume the push again. I don't really know of another
solution to our current problem.
josh