[CALCITE-4896] GROUP BY might get wrong results under certain circumstances in Elasticsearch Adapter

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.28.0
• Component/s: elasticsearch-adapter
• Assignee: ZheHu
• Reporter: ZheHu
• Created: 20/Nov/21 01:20
• Updated: 17/Dec/21 00:57

Description

In Elasticsearch Adapter, if one field is grouped by along with missing value, SQL user might get incorrect results. Take the following case as an example:

ES mappings: {"int_field1" : integer, "int_field2" : integer}

doc1 = {"int_field1":1, "int_field2": -2147483648}
doc1 = {"int_field1":2}

When I try "select count(1) as CNT from view group by int_field2", the result is 2.

For Integer type, the missing value is replaced by Integer.MIN_VALUE, so doc1 and doc2 will be divided in the same group.

Any other data types like short、long、float、double, they also have such problem.

Issue Links

• Blocked: CALCITE-2689 ES Adapter. Grouping on date / number fields fails