PSMDB-2163 vector search

docs/vector_search/configure-mongot.md

@@ -0,0 +1,29 @@
+# Configure mongot
+
+After installing `mongot`, configure it to connect to the Percona Server for MongoDB replica set, store search indexes, and accept search requests from `mongod`.
+
+Vector Search requires configuration on both sides:
+
+- `mongot` must be able to connect to the replica set and synchronize data.
+- `mongod` must know where to send search queries and search-index management requests.
+
+!!! note
+    The examples in this section use a single-node test environment with `mongod` listening on port **27017** and `mongot` listening on port **27028**. Replace the host names, ports, paths, and credentials with values that are applicable for your deployment.
+
+## Before you begin
+
+- PSMDB is running as an initialized replica set.
+- Authentication is enabled.
+- The `mongot` binary is installed.
+- The user running `mongot` can write to the search data and log directories.
+- The host running `mongot` can connect to the replica set.
+- The required ports are available.
+
+## Procedure
+
+
+
+
+
+
+

docs/vector_search/create-search-index.md

@@ -0,0 +1,167 @@
+# Create Index
+
+You can use the `createSearchIndex()` method to create a single Search or Vector Search index on a collection, or the `createSearchIndexes()` method to create multiple indexes simultaneously.
+
+## Create single Search Index
+
+Follow these steps to create a single search index:
+{.power-number}
+
+1. Insert test data from mongosh:
+
+    ```javascript
+    use test
+    db.docs.insertMany([
+    { text: "MongoDB search is powerful" },
+    { text: "Vector search is the future" },
+    { text: "Full text search with mongot" }
+    ])
+
+    {
+    acknowledged: true,
+    insertedIds: {
+        '0': ObjectId('69ebae2599c54be2ea44ba89'),
+        '1': ObjectId('69ebae2599c54be2ea44ba8a'),
+        '2': ObjectId('69ebae2599c54be2ea44ba8b')
+    }
+    }
+    ```
+
+2. Create a **single Search Index**:
+
+    ```javascript
+    db.docs.createSearchIndex({
+    name: "search_idx",
+    definition: {
+        mappings: {
+        dynamic: true
+        }
+    }
+    })
+
+    search_idx
+    ```
+
+3. Check the status.
+
+    ```javascript
+    db.docs.getSearchIndexes()
+    [
+    {
+        id: '69ebae2a651bce4d10f57bdc',
+        name: 'search_idx',
+        status: 'READY',
+        queryable: true,
+        latestDefinitionVersion: { version: 0, createdAt: ISODate('2026-04-24T17:53:46.000Z') },
+        latestDefinition: { mappings: { dynamic: true } },
+        statusDetail: [
+        {
+            hostname: '69eb5fc573906b6bfb8cefe7',
+            status: 'READY',
+            queryable: true,
+            mainIndex: {
+            status: 'READY',
+            queryable: true,
+            definitionVersion: { version: 0, createdAt: ISODate('2026-04-24T17:53:46.000Z') },
+            definition: { mappings: { dynamic: true, fields: {} } }
+            }
+        }
+        ]
+    }
+    ]
+    ```
+
+## Create multiple Search Indexes
+
+The following example creates two Search indexes:
+
+```javascript
+db.docs.createSearchIndexes([
+  {
+    name: "search_idx",
+    definition: {
+      mappings: {
+        dynamic: true
+      }
+    }
+  },
+  {
+    name: "text_search_idx",
+    definition: {
+      mappings: {
+        dynamic: false,
+        fields: {
+          text: {
+            type: "string"
+          }
+        }
+      }
+    }
+  }
+])
+[ 'search_idx', 'text_search_idx' ]
+```
+This creates:
+
+- `search_idx`, which dynamically indexes supported fields.
+- `text_search_idx`, which indexes only the text field.
+
+## Create Search and Vector Search Indexes together
+
+You can include Search and Vector Search index definitions in the same command.
+
+```javascript
+db.docs.createSearchIndexes([
+  {
+    name: "search_idx",
+    type: "search",
+    definition: {
+      mappings: {
+        dynamic: false,
+        fields: {
+          text: {
+            type: "string"
+          }
+        }
+      }
+    }
+  },
+  {
+    name: "vector_idx",
+    type: "vectorSearch",
+    definition: {
+      fields: [
+        {
+          type: "vector",
+          path: "embedding",
+          numDimensions: 3,
+          similarity: "cosine"
+        }
+      ]
+    }
+  }
+])
+[ 'search_idx', 'vector_idx' ]
+```
+
+## Supported field types
+
+The following field types are supported:
+
+| Field type | Supported source data| Use|
+| -----------| ----------------------| --|
+| `boolean`           | Boolean                       | Supports exact matching and filtering on `true` and `false` values.                                |
+| `date`              | BSON Date                     | Supports exact matches, range and proximity queries, sorting, and faceting.                        |
+| `document`          | Object or subdocument         | Defines mappings for fields within a nested object.                                                |
+| `embeddedDocuments` | Array of objects              | Indexes objects within an array so that each embedded object can be evaluated independently.       |
+| `number`            | `int32`, `int64`, or `double` | Supports exact matches, numeric range and proximity queries, sorting, and faceting.                |
+| `objectId`          | `ObjectId`                    | Supports matching, filtering, and range operations on `ObjectId` values.                           |
+| `string`            | String                        | Analyzes text for full-text operators such as `text`, `phrase`, `regex`, and `wildcard`. |
+
+For detailed inforamtion on supported feild types, see MongoDB documentation](https://www.mongodb.com/docs/search/index/define-field-mappings/#mongodb-search-field-types)
+
+
+[Update a search index :material-arrow-right:](../update-search-index.md){.md-button}
+
+[Delete a search index :material-arrow-right:](../delete-search-index.md){.md-button}
+

docs/vector_search/delete-search-index.md

@@ -0,0 +1,14 @@
+# Delete Index
+
+Use the `dropSearchIndex()` method to delete a search or vector search index. The method works the same way for both index types:
+
+```javascript
+db.products.dropSearchIndex("products_text_idx");
+```
+
+Deleting an index is irreversible. To restore search functionality, create the index again with `createSearchIndex()`, and wait for the initial sync to
+complete before running queries against it.
+
+!!! warning
+    `$search` and `$vectorSearch` queries that reference an index that does not exist return an empty result set rather than an error. If your application starts receiving empty search results after an index change, verify that the index exists and check its status with `getSearchIndexes()`.
+