Clarify the differences between the three options for repairing broken repositories

[quba42]

URL of the document

https://docs.theforeman.org/nightly/Managing_Content/index-katello.html#Recovering_a_Corrupted_Repository_content-management

Sections 5.26 to 5.30

Describe the issue

There are broadly three different actions one can take to repair different kind of repo issues:

1. Complete Sync
2. Verify Content Checksum
3. Republish repository metadata

Variants of these actions are available for the server as well as the proxies, and they can be scoped to individual repos, products, or content views. The current version of the docs could be clearer on which of these actions needs to be used for what kinds of issues, and contains some mistakes in the details.

Anything else to add? (context, suggestions for a fix, etc.)

Example issues in the docs:

• 5.26: Verify Content Checksum does not "Synchronizes all packages and then verifies the checksum", it verifies the checksum and then only re-downloads those packages where the checksum did not match. It is also not true that this only applies to yum content. In fact since the underlying Pulp feature operates on the level of artifacts it should work for all content types.
• 5.26: Provides an overview of Optimized Sync, Complete Sync, Verify Content Checksum, but not Republish repository metadata
• 5.29: Docs warn against using this feature. I believe the background is issues around RPM full mirror mode. It should not be discouraged for other content types (I can say with expert knowledge that this is a pretty safe workflow for deb content).
• 5.29: "Foreman community recommends a complete repository sync", if the complete sync does not result in new content, then this will not trigger a re-publish either, so this is not a good alternative to a re-publish.