Double aggregation with unique documents

[Ruben_Fanjul]

Tables

Orders
  orderstate
  time_reserved
  user_id
  product_id

Product
  id
  refid

This query hit 76 documents.

select * from Orders
	inner join Products on Orders.productid = Products.id 
	where Products.`refid` = 9
	and orderstate like "confirmed"
    group by userid; 

With the list of all unique users and I would like to group that with time_reserved to know the list of unique users first and then group this list by "day" counting the users in this day.

My ElasticSearch query is:

{
	"aggs": {
		"group_by_date": {
			"date_histogram": {
				"field": "time_reserved",
				"interval": "day",
				"format": "yyyy-MM-dd"
			},
			"aggs": {
				"amount": {
					"cardinality": {
						"field": "user_id"
					}
				}
			}
		}
	},
	"query": {
		"bool": {
			"filter": [{
				"term": {
					"dealid": "9"
				}
			}],
			"must": [{
				"terms": {
					"orderstate": ["confirmed"]
				}
			}]
		}
	}
}

Which works but it hits 144 documents, that is because it is not doing the cardinality in the second aggregation and it is returning duplicate user_id documents and match with the result of the same query above but without group by:

select * from Orders
  inner join Products on Orders.productid = Products.id 
  where Products.`refid` = 9
  and orderstate like "confirmed"; 

There is a simple way to have to group by ?