Goal

While ideally repositories should not have duplicate packages in them, they often do. Yum/Dnf do a great job of not letting duplicate packages complicate things when installing and resolving dependencies. However, the package count shown by yum/dnf does include the package count of all packages (including duplicates). When comparing package counts from a RHEL client to the repodata created by Pulp on Red Hat Satellite or from Red Hat CDN, repository package count numbers can be the first option for users to verify their client is seeing everything available to the client from their repository source.

Pulp on the CDN and Satellite already exclude duplicate packages from their total package count when generating metadata for the repository, but yum/dnf don't. By including the duplicate packages from a repository the client will show more packages available than the repository source declares are available. While yum/dnf are technically more accurate in this situation, they do nothing except cause confusion for a user when clients see more packages than their sources show.

While Red Hat could fix the duplicate packages in a repository when they are present, we can't control 3rd party repositories where they could also be present. For this reason, I'm filing the RFE for yum/dnf to adopt the same policy of de-duplicating packages in its repository package count and use the package count provided by the primary.xml files in the repodata.

Previous examples of this issue:

https://issues.redhat.com/browse/RHELDST-29405 

https://issues.redhat.com/browse/RHELDST-27978 

https://issues.redhat.com/browse/RHELDST-18874 

Acceptance criteria

A list of verification conditions, successful functional tests, or expected outcomes in order to declare this story/task successfully completed.

• While duplicate packages exist in a repository, `dnf repolist -v` shows a package count matching that of the primary.xml from the repodata.