Currently hashed indexes are not allowed to be unique (due to potential hash collisions on the keys). Uniqueness can be enforced on a field if a non-hashed index is created on the same field, but this is inefficient.
It should be possible to create a unique hashed index by checking for a match on the actual document (not just on the key in the index) before insertion.
- has to be done before
-
SERVER-8031 Allow dropping the _id index when you have a hashed _id index
-
- Closed
-
- is related to
-
-
- Closed
-