> For the complete documentation index, see [llms.txt](https://qyverlabs.gitbook.io/qyverlabs-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://qyverlabs.gitbook.io/qyverlabs-docs/run-in-production/supported-vector-databases/mongo-db.md).

# Mongo DB

#### Integrating MongoDB with Qyver

This guide provides step-by-step instructions on configuring and integrating MongoDB with Qyver.

**Configuring an Existing Managed MongoDB Instance**

To use MongoDB with Qyver, ensure your MongoDB version supports **Atlas Vector Search**. Refer to the [official MongoDB documentation](https://www.mongodb.com/docs/atlas/atlas-vector-search/vector-search-overview/) for more details.

Qyver requires access to MongoDB to manage **Atlas Search Indexes**. MongoDB’s instance size affects how these indexes can be created, listed, and deleted. Specifically:

* Instances below **M10** do not allow these operations via a standard user and require access through the **administration API**.
* Larger instances allow direct index management.

To support all instance types, Qyver uses the **administration API** for index management. This requires an **API key** with the **Project Data Access Admin** role. Instructions for creating this key are included below.

**Note:** The API requires both a **project ID** and **cluster name**. Instructions on retrieving this information are also provided.

***

#### Configuration Modifications

To integrate MongoDB, add the `MongoDBVectorDatabase` class and include it in the executor:

```python
from qyver import framework as qv

vector_database = qv.MongoDBVectorDatabase(
    host="<USER>:<PASSWORD>@<HOST_URL>",  # MongoDB connection string, including username and password
    db_name="<DATABASE_NAME>",  # Name of the database (must be created manually)
    cluster_name="<CLUSTER_NAME>",  # Name of the cluster in MongoDB Atlas
    project_id="<PROJECT_ID>",  # Project ID (not the project name)
    admin_api_user="<API_USER>",  # API key user (public key in MongoDB Atlas)
    admin_api_password="<API_PASSWORD>",  # API password (private key in MongoDB Atlas)
    default_query_limit=10,  # (Optional) Maximum number of results per query (defaults to 10)
    # Additional parameters can be passed as kwargs and will be handled by the MongoDB client.
)
```

**Finding Your Project ID**

* In MongoDB Atlas, select your **organization** from the top left corner.
* Find your **project** but do not click on it.
* Click the **Actions (···)** menu in the last column and select **Copy Project ID**.

Alternatively, your **Project ID** is in the URL when viewing a project:\
`https://cloud.mongodb.com/v2/<PROJECT_ID>#/overview`\
Make sure to copy the **correct ID** and not the organization ID.

**Setting Up the Executor**

Once the MongoDBVectorDatabase is configured, assign it to the executor:

```python
...
executor = qv.RestExecutor(
    sources=[source],
    indices=[index],
    queries=[qv.RestQuery(qv.RestDescriptor("query"), query)],
    vector_database=vector_database,  # Use the MongoDBVectorDatabase instance
)
...
```

***

#### Starting a Managed MongoDB Instance

Follow these steps to set up a **MongoDB Atlas** instance:

1. **Create the Database**
   * Sign in to MongoDB Atlas.
   * Create a cluster and note the **cluster name**.
   * Click **Database** from the left menu.
   * Once the cluster is created, open it and navigate to the **Collections** tab.
   * Click **Add My Own Data**, enter a database name, and create a collection.
2. **Create a Database User and Allow IP Access**
   * Click **Database** from the left menu.
   * Click **Connect** next to your cluster.
   * In the pop-up, allow access from **any IP** or specify your machine’s IP.
   * Create a **database user** with a username and password.
3. **Generate an API Key**
   * Go to **Access Manager** and select your **project**.
   * Open the **API Keys** tab.
   * Create an API key with the **Project Data Access Admin** role.
   * Save the **private key** (it will not be accessible again).