createSearchIndexes

On this page

Definition

Syntax
Command Fields
Search Index Definition Syntax
Vector Search Index Definition Syntax
Behavior
Access Control
Output
Examples
Create a Search Index on All Fields
Create a Search Index with a Language Analyzer
Create Multiple Search Indexes
Create a Vector Search Index
Learn More

Definition

createSearchIndexes

New in version 7.0: (Also available starting in 6.0.7)

Creates one or more Atlas Search indexes or Vector Search indexes on a specified collection.

The mongosh method db.collection.createSearchIndex() provides a wrapper around the createSearchIndexes database command.

Important

This command can only be run on a deployment hosted on MongoDB Atlas, and requires an Atlas cluster tier of at least M10.

Syntax

Command syntax:

db.runCommand(
   {
      createSearchIndexes: "<collection name>",
      indexes: [
         {
            name: "<index name>",
            type: "<search index type>",
            definition: {
               /* search index definition fields */
            }
        },
        ...
      ]
   }
)

Command Fields

The createSearchIndexes command takes the following fields:

Field	Type	Necessity	Description
`createSearchIndexes`	string	Required	Name of the collection on which to create the search index.
`indexes`	array	Required	Array of documents describing the indexes to create.
`indexes.name`	string	Optional	Name of the search index to create. You cannot create multiple indexes with the same name on a single collection. If you do not specify a `name`, the index is named `default`.
`indexes.type`	string	Optional	Type of search index to create. You can specify either: `search` `vectorSearch` If you omit the `type` field, the index type is `search`.
`indexes.definition`	document	Required	Document describing the index to create. The `definition` syntax depends on whether you create a standard search index or a Vector Search index. For the `definition` syntax, see: Search Index Definition Syntax Vector Search Index Definition Syntax

Search Index Definition Syntax

The search index definition takes the following fields:

{
   analyzer: "<analyzer-for-index>",
   searchAnalyzer: "<analyzer-for-query>",
   mappings: {
      dynamic: <boolean>,
      fields: { <field-definition> }
   },
   analyzers: [ <custom-analyzer> ],
   storedSource: <boolean> | {
      <stored-source-definition>
   },
   synonyms: [ {
      name: "<synonym-mapping-name>",
      source: {
         collection: "<source-collection-name>"
      },
      analyzer: "<synonym-mapping-analyzer>"
   } ]
}

Field	Type	Necessity	Description
`analyzer`	string	Optional	Specifies the analyzer to apply to string fields when indexing. If you omit this field, the index uses the standard analyzer.
`searchAnalyzer`	string	Optional	Specifies the analyzer to apply to query text before the text is searched. If you omit this field, the index uses the same analyzer specified in the `analyzer` field. If you omit both the `searchAnalyzer` and the `analyzer` fields, the index uses the standard analyzer.
`mappings`	object	Optional	Specifies how to index fields on different paths for this index.
`mappings.dynamic`	boolean	Optional	Enables or disables dynamic field mapping for this index. If set to `true`, the index contains all fields containing supported data types. If set to `false`, you must specify individual fields to index using `mappings.fields`. If omitted, defaults to `false`.
`mappings.fields`	document	Conditional	Required only if dynamic mapping is disabled. Specifies the fields to index. To learn more, see Define Field Mappings.
`analyzers`	array	Optional	Specifies the Custom Analyzers to use in this index.
`storedSource`	boolean or Stored Source Definition	Optional	Specifies document fields to store for queries performed using the returnedStoredSource option. You can store fields of all Data Types on Atlas Search. The `storedSource` value can be one of these: `true`, to store all fields `false`, to not store any fields An object that specifies the fields to `include` or `exclude` from storage If omitted, defaults to `false`. To learn more, see Define Stored Source Fields in Your Atlas Search Index.
`synonyms`	array of Synonym Mapping Definitions	Optional	Specifies synonym mappings to use in your index. Configuring synonyms allows you to you index and search for words that have the same or a similar meaning. To learn more, see Define Synonym Mappings in Your Atlas Search Index.

Vector Search Index Definition Syntax

The vector search index definition takes the following fields:

{
  "fields": [
    {
      "type": "vector" | "filter",
      "path": "<field-to-index>",
      "numDimensions": <number-of-dimensions>,
      "similarity": "euclidean" | "cosine" | "dotProduct"
    }
  ]
}

For explanations of vector search index definition fields, see How to Index Fields for Vector Search.

Behavior

The createSearchIndexes command triggers an index build. There may be a delay between when you receive a response from the command and when the index is ready.

To see the status of your search indexes, use the $listSearchIndexes aggregation stage.

Access Control

If your deployment enforces access control, the user running the createSearchIndexes command must have the createSearchIndexes privilege action on the database or collection:

{
   resource: {
      db : <database>,
      collection: <collection>
   },
   actions: [ "createSearchIndexes" ]
}

The built-in readWrite role provides the createSearchIndexes privilege. The following example grants accountUser01 the readWrite role on the products database:

db.grantRolesToUser(
   "accountUser01",
   [ { role: "readWrite", db: "products" } ]
)

Output

The createSearchIndexes command output resembles the following:

{
   ok: 1,
   indexesCreated: [
      {
         id: "<index Id>",
         name: "<index name>"
      }
   ]
}

Important

The response field ok: 1 indicates that the command was successful. However, there may be a delay between when you receive the response and when the created indexes are ready for use.

To see the status of your search indexes, use the $listSearchIndexes aggregation stage.

Examples

Create a Search Index on All Fields

The following example creates a search index named searchIndex01 on the contacts collection:

db.runCommand( {
   createSearchIndexes: "contacts",
   indexes: [
      {
         name: "searchIndex01",
         definition: { mappings: { dynamic: true } }
      }
   ]
} )

The index definition specifies mappings: { dynamic: true }, which means that the index contains all fields in the collection that have supported data types.

Create a Search Index with a Language Analyzer

A language analyzer introduces stop-words, which are words that are not significant enough to be indexed.

The following example creates a search index named frenchIndex01 on the cars collection, and specifies the lucene.french analyzer on the fr field:

db.runCommand( {
   createSearchIndexes: "cars",
   indexes: [
      {
         name: "frenchIndex01",
         definition: {
            mappings: {
              fields: {
                subject: {
                  fields: {
                    fr: {
                      analyzer: "lucene.french",
                      type: "string"
                    }
                  },
                  type: "document"
                }
              }
            }
          }
      }
   ]
} )

To learn more about language analyzers, see Language Analyzers.

Create Multiple Search Indexes

The following command creates two search indexes on the products collection, searchIndex02 and searchIndex03:

db.runCommand( {
   createSearchIndexes: "products",
   indexes: [
      {
         name: "searchIndex02",
         definition: {
            mappings: {
               fields: {
                  title: {
                     type: "string",
                     analyzer: "lucene.simple"
                  }
               }
            }
         }
      },
      {
         name: "searchIndex03",
         definition:
            {
               mappings: { dynamic: true }
            }
      }
   ]
} )

searchIndex02 uses a simple analyzer on the title field. The simple analyzer divides text into searchable terms based on non-letter characters, such as whitespace, punctuation, or digits.

searchIndex03 uses a dynamic field mapping, meaning the index contains all fields in the collection that have supported data types.

Create a Vector Search Index

The following example creates a vector search index named vectorSearchIndex01 on the movies collection:

db.runCommand( {
   createSearchIndexes: "movies",
   indexes: [
      {
         name: "vectorSearchIndex01",
         type: "vectorSearch",
         definition: {
            fields: [
               {
                  type: "vector",
                  numDimensions: 1,
                  path: "genre",
                  similarity: "cosine"
               }
            ]
         }
      }
   ]
} )

The vector search index contains one dimension and indexes the genre field.

Learn More

Back

Atlas Search

dropSearchIndex