When performing foreground validation with repair:false, the default for standalone nodes is to run the validation with the repair mode set to kAdjustMultikey. In this mode, mutlikey index inconsistencies related to a multikey index being too permissive (e.g, an index being labelled multikey when none of the documents it covers have array fields anymore) are fixed up, but no correctness issues are resolved.

However, it can be confusing to see messages regarding an index being updated in the "warnings" section of the validate results when validation is run in repair:false mode. Also, when multikeys are adjusted in this way the "repaired" field of the validate results object is set to true, which can be confusing (validation was run with repair:false but it looks like the collection was repaired).

This ticket would look into clarifying the log message and seeing if there's a way to make the behavior less confusing from the customer perspective.