Definition of index-templates for whole sub-structures needed?

[ppuschmann (Paul Puschmann)]

Hi,
we're currently having an interesting issue with our field-definitions and type-detection.

We have fields in our documents that sometimes contain integers, sometimes some strings.

In the documentation we find some parts regarding object datatypes: https://www.elastic.co/guide/en/elasticsearch/reference/5.6/object.html

Given we have this template defined:

PUT my_index
{
  "mappings": {
    "my_type": { 
      "properties": {
        "region": {
          "type": "keyword"
        },
        "manager": { 
          "properties": {
            "age":  { "type": "integer" },
            "name": { 
              "properties": {
                "first":     { "type": "text" },
                "last":      { "type": "text" },
                "somefield": { "type": "keyword" }
              }
            }
          }
        }
      }
    }
  }
}

The field somefield is in this case our field with mixed input types.

Would this index-template be valid for documents, that don't have a field somefield (or name.somefield)?

So, is the example-template valid for all four example-documents?

Example 1:

PUT my_index/my_type/1
{ 
  "region": "US",
  "manager": { 
    "age":     30,
    "name": { 
      "first": "John",
      "last":  "Doe"
    }
  }
}

Example 2:

PUT my_index/my_type/2
{ 
  "region": "US",
  "manager": { 
    "age":     35,
    "name": { 
      "first":     "Patty",
      "last":      "Myers",
      "somefield": "Some text..."
    }
  }
}

Example 3:

PUT my_index/my_type/3
{ 
  "region": "US",
  "manager": { 
    "age":     35,
    "name": { 
      "first":     "Patrick",
      "last":      "McDonald",
      "somefield": 42
    }
  }
}

Example 4:

PUT my_index/my_type/4
{ 
  "region": "US",
  "manager": { 
    "age":     35,
    "name": { 
      "first":              "Patrick",
      "last":               "McDonald",
      "somefield":          42,
      "somedifferentfield": 42
    }
  }
}

[abdon (Abdon Pijpelink)]

Yes, fields are optional in Elasticsearch, so an index with your mapping will happily accept those documents. Any unmapped field that shows up in a document (like your somedifferentfield) will be dynamically mapped, in your case to a long type.

[ppuschmann (Paul Puschmann)]

Thanks for the clarification.