MongoDB connection

Connect a task to MongoDB via a managed Leoflow Connection and MongoHook. The conn_type is mongo. It is the host:port + login/password shape; the Schema field carries the database (or auth database).

Declare the provider

# leoflow.yaml
dag_id: mongo_load
connectors:
  - mongo

Fields the UI asks for

Field	Notes
Host	Mongo host (or the first of a replica set).
Port	Defaults to 27017.
Login / Password	The Mongo user. Password encrypted at rest (ADR 0019).
Schema	The database, e.g. analytics.
Extra	{"srv": true} for mongodb+srv:// (Atlas), {"authSource": "admin"}, TLS options.

The host:port + login/password + db round-trip is covered by the table-driven TestConnectionDeliveryChainOfCustodyIntegration (mongo row).

Example DAG (copy-paste)

Docs-only recipe (needs a MongoDB instance). Hook imported inside the task.

# dag.py
from __future__ import annotations

from airflow.sdk import DAG, task


@task
def load() -> None:
    from airflow.providers.mongo.hooks.mongo import MongoHook

    hook = MongoHook(mongo_conn_id="mongo_default")
    print("load: connecting via MongoHook(mongo_default)")
    coll = hook.get_collection("example_load", mongo_db="analytics")
    coll.delete_many({})
    coll.insert_many([{"name": f"cat_{i}", "score": (i * 7) % 100} for i in range(20)])
    print(f"load: {coll.count_documents({})} docs in example_load")


with DAG("mongo_load", schedule=None, catchup=False, tags=["example"]):
    load()

# leoflow.yaml
schema_version: "1.0"
dag_id: mongo_load
description: Load documents into MongoDB via MongoHook.
owner: examples
tags: [example]
python_version: "3.11"
connectors:
  - mongo

Atlas (mongodb+srv)

For MongoDB Atlas set Extra: {"srv": true} and use the cluster host (cluster0.xxxx.mongodb.net) without a port — MongoHook builds the mongodb+srv:// URI.

Related

• docs/connections/index.md — Installing a connector's provider.
• ADR 0019 / 0021 — secret encryption + agent delivery.