What you’ll need
- A MongoDB Atlas cluster or self-managed MongoDB deployment reachable from Summation.
- The database name and either a host or a MongoDB connection string.
- If authentication is enabled, a database user that can read the selected collections and run
listCollections. MongoDB’s built-inreadrole is a common starting point. - For Atlas, an IP access list entry that permits connections from Summation.
For MongoDB Atlas, enter only the SRV hostname, such as
cluster0.abc123.mongodb.net. Do not include mongodb+srv:// in the Host field. Summation detects Atlas hostnames and uses SRV discovery automatically.Form fields
These fields are available in the connection wizard.Connect to MongoDB Atlas
1
Allow network access
Add Summation’s egress addresses to the Atlas project’s IP access list.
2
Create a database user
Create a user with read access to the database and collections you want to expose.
3
Enter the SRV hostname
Copy the hostname from the Atlas connection string. For
mongodb+srv://cluster0.abc123.mongodb.net/, enter cluster0.abc123.mongodb.net as Host. Keep TLS mode set to required.4
Test and select collections
Click Test connection, then select the collections Summation should add as datasets.
Advanced configuration
The connector API accepts these additional configuration fields. They are not currently shown in the connection wizard.mongodb_connection_string and mongodb_pass are secrets. Send them in the connector’s secrets payload, never in its non-secret config. Saved secret values are write-only and are not returned by the API.Schema inference and nested documents
MongoDB collections can contain documents with different fields and types. Summation samplesmongodb_num_docs_to_infer_schema documents to establish the dataset schema.
Increase the sample count when fields are sparse or appear only in later documents. Larger samples improve coverage but make initial schema discovery slower.
Enable Flatten nested objects to expose the first level of embedded documents as columns. For example, customer.region becomes a column while deeper objects remain JSON strings. Arrays remain array values rather than being expanded into multiple rows. Leave the setting off to keep embedded documents as JSON values.