If you have an index on a:1, b:1 and distinct command or agg equivalent
db.foo.distinct("b", {a:1})
we use DISTINCT_SCAN unless b is multikey - and it's not clear why - it doesn't need IXSCAN nor FETCH (which is what it shows using).
- is duplicated by
- 
                    SERVER-27494 Avoid unwind on multi-key index field for covered index aggregation -         
- Backlog
 
-