This is a tracking-only bug for software assuming rpmdb being in Berkeley DB format, such as testing for particular file names that are not valid with other rpm database backends (alternatives with differing filenames have been available since rpm 4.13)
These assumptions have surfaced while testing ndb on c8s. I am planning on following the various dependencies listed here, and producing a patch set that works for c8s. We'll do validation and stress testing. We are keen to get off bdb in c8s like we did in c7.
Berkeley DB is now gone from rawhide, and wont be coming back.