Connectors

A connector is a reusable connection configuration for an external data source. It stores connectivity information such as addresses, credentials, authentication, and timeouts, and encapsulates read/write capabilities for reuse by data loading, data export, queries, knowledge bases, and other features.

A connector defines only how to connect. For the complete list of 9 categories / 41 data source types and all connection fields, see Reference · Connectors. This page covers only creation, testing, maintenance, and permissions.

List and purposes

Go to Data Connection > Connectors to view connectors accessible in the current workspace. The list typically shows the name, data source category, specific data source, purpose, creation time, and actions such as permissions, edit, and delete. You can search by name, category, or data source.

Select at least one purpose when creating a connector:

Purpose

Data flow

Tasks that can use it

Load

External → MOI

Data loading, external data reads, knowledge base data access, and more

Export

MOI → external

Data export tasks

Load + export

Bidirectional

The same data source serves as both a source and an export destination

The purpose determines whether the connector appears in the corresponding task selector. Before changing it, confirm that no running task will lose a required capability.

Create and test

  1. On the connector list, click Create connector.

  2. Enter a name that is unique within the workspace and identifies the system and purpose, such as “Production MySQL - orders read-only.”

  3. Select a data source category and a specific data source.

  4. Select Load, Export, or both.

  5. Enter the connection information for the selected type. See Reference · Connectors for field definitions; fields marked with an asterisk in the UI are required. Use a least-privilege account for credentials.

  6. Click Test connection. Create the connector only after the test succeeds. If it fails, check the address, port, network, authentication, and source-side authorization according to the error.

  7. Click Create to save.

Recommendation A successful connection test only means that the current parameters can establish a connection. Before the first load or export, still use a small data set to verify the source path, database and table permissions, API response format, and target write permissions.

Maintenance and permissions

Action

Description

Edit

Change the name, purpose, or connection parameters; retest after credentials, networking, or authentication changes

Test connection

Distinguish an unreachable data source from a task configuration or data problem

Delete

Before deleting, check load and export tasks that reference the connector; afterward, those tasks can no longer access the source

Permissions

Control who can view or use each connector to avoid unnecessarily exposing production credentials

Deleting a connector does not automatically repair tasks that reference it. Replace it in load tasks or export tasks before deleting it.

Boundaries to remember during configuration

See the reference for the complete field tables. During operation, remember only these boundaries:

  • Databases: A connector usually stops at the instance level; select the database, table, or collection during loading or querying. Oracle Service Name / SID identifies the connection target itself and must be entered in the connector.

  • Files and object storage: The connector stores the entry point and root path; select specific files or folders during loading.

  • Message queues: In addition to network and authentication settings, fields must define consumption semantics such as the topic or queue, consumer group, starting offset, or QoS.

  • Email: The connector manages authorization; configure the time range and folder in the load task.

  • REST / GraphQL: The connector manages the Base URL / endpoint and authentication; configure the specific endpoint, method, and parameters in the load task.

If external database catalogs / federated queries are available, they should reuse an existing database connector. Whether data is materialized in MOI and whether queries are pushed down depends on the subsequent use, not on connector fields themselves.

Troubleshooting order

  1. Run Test connection on the connector again.

  2. Verify that the source account still has access to the target bucket, path, database table, topic, API, or business object.

  3. Confirm that the database table, path, or endpoint selected by the task matches the connector configuration.

  4. Review runtime errors in the load task or export task details.

If a field is unclear, open Reference · Connectors and consult the table for that data source.