An odd requirement has come my way. One of our indexes has uniqueness on two different fields, but because Solr only allows one uniqueKey field, we cannot have automatic document replacement on both of the fields. This means that the indexing code must handle it, which (for reasons I don't fully understand) currently results in some *massive* delete requests being sent frequently -- one such request was over 130KB in size. They look like "field:(x || y || z)" -- but with a LOT of different values.
How much pain would it take to implement multiple uniqueKeys? I have not searched Jira for an existing issue. Thanks, Shawn