> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mantrixflow.com/llms.txt
> Use this file to discover all available pages before exploring further.

# MongoDB and PostgreSQL verified paths

> Build and verify MongoDB-to-PostgreSQL, PostgreSQL-to-MongoDB, and MongoDB-to-MongoDB example pipelines.

This guide summarizes the three MongoDB data-movement paths verified against a
real standalone MongoDB 8.0 deployment and PostgreSQL 17.

<Note>
  Example records and identifiers in this guide are synthetic test fixtures,
  not customer data or customer activity.
</Note>

| Source     | Destination                  | Verification                                            |
| ---------- | ---------------------------- | ------------------------------------------------------- |
| MongoDB    | PostgreSQL                   | Passed with an Incremental rerun and no duplicates      |
| PostgreSQL | MongoDB                      | Passed with an updated source row and idempotent Upsert |
| MongoDB    | Different MongoDB collection | Passed with `_id` Upsert and no internal dlt fields     |

Other SQL pairings, Atlas, replica sets, sharded clusters, and live TLS have
pending deployment-specific validation. Available Now does not imply that each
environment has been independently certified.

## Prerequisites

1. Create a [MongoDB source connection](/connections/sources/database/mongodb).
2. Create a [MongoDB destination connection](/connections/destinations/mongodb)
   when testing delivery into MongoDB.
3. Create PostgreSQL reader or writer connections as required.
4. Use separate MongoDB source and destination collections.
5. Prepare stable keys and supporting indexes before repeated runs.

## MongoDB to PostgreSQL

1. Create a pipeline with the MongoDB source.
2. Discover the required `database.collection` stream.
3. Choose **Incremental** and select a stable cursor such as `updated_at`.
4. Add the PostgreSQL destination and publish a SQL model that includes `_id`,
   the cursor, and the required business fields.
5. Map the PostgreSQL destination table and select `_id` as the Upsert key.
6. Run the initial load.
7. Add or update a MongoDB document with a later cursor value and run again.

Verify that the PostgreSQL row count increases only for new keys and that tied
cursor-boundary documents do not create duplicates.

```sql theme={"theme":{"light":"github-light","dark":"github-dark"}}
SELECT COUNT(*) AS delivered_rows
FROM public.mongo_orders;

SELECT _id, COUNT(*) AS occurrences
FROM public.mongo_orders
GROUP BY _id
HAVING COUNT(*) > 1;
```

## PostgreSQL to MongoDB

1. Create a pipeline with the PostgreSQL source.
2. Select the source table and configure Full Table or Incremental extraction.
3. Add the MongoDB destination.
4. Publish a SQL model containing a stable business key such as `id`.
5. Set the output target to `analytics.pg_orders` and select `id` as the Upsert
   key.
6. Run the pipeline, update one PostgreSQL row, and run again.

Confirm that the collection count is unchanged after the update rerun and the
matching document contains the new value:

```javascript theme={"theme":{"light":"github-light","dark":"github-dark"}}
use analytics

db.pg_orders.countDocuments({})
db.pg_orders.findOne({ id: "pg-1" })
```

## MongoDB to MongoDB

Use different source and destination collections. A pipeline targeting the same
logical database and collection is rejected.

1. Select the source collection and use `_id` as the stable record identity.
2. Add the destination connection and choose another database or collection.
3. Publish the transformed document fields.
4. Use `_id` as the Upsert key.
5. Run twice and verify that the destination count remains stable.

```javascript theme={"theme":{"light":"github-light","dark":"github-dark"}}
use copy_sink

db.events_copy.countDocuments({})
db.events_copy.find({ _dlt_id: { $exists: true } })
```

The second query should return no documents. MantrixFlow removes all top-level
`_dlt*` fields before MongoDB delivery.

## Operational checks

* Test every connection before saving it.
* Keep source identities read-only and destination identities scoped to one
  database.
* Index source cursor fields and destination Upsert keys.
* Compare run row counts with direct database counts.
* Confirm a failed run does not advance the previous successful checkpoint.
* Review MongoDB index warnings before increasing batch volume.
* Never paste a complete MongoDB URI into SQL, logs, tickets, or screenshots.

For connector details and current limitations, read the
[MongoDB source](/connections/sources/database/mongodb) and
[MongoDB destination](/connections/destinations/mongodb) references.