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

# Connect Microsoft Fabric

> Connect a Fabric workspace, prepare supported tables, and manage the workspace access used by your execution environment.

Connect a Fabric workspace in **Discovery** to browse supported Lakehouse and
Warehouse tables and import the ones you want to validate. Your customer
execution Data Factory discovers the table metadata and checks access; source
rows remain in your environment.

## Before you connect

* Complete [environment setup](/installing-general-validation) and wait for
  the execution environment to become ready.
* Use an **Owner** account in General Validation that is also an **Admin** or
  **Member** of the Fabric workspace.
* Ask your Fabric administrator to allow the customer execution identity to
  use the required Fabric APIs. Tenant policy may also require an Entra
  administrator to approve the browser's Fabric permission.
* Confirm that Fabric capacity is active and that the execution factory can
  reach the supported SQL endpoints.

Review [Supported data and tests](/supported-scope) for the supported table
types, schemas, and validation limits.

## Review workspace access

Guided setup requests **Viewer** access for the managed identity of your
customer execution Data Factory. Viewer permits reads across the selected
workspace, including Lakehouse and Warehouse SQL data. Review that scope with
your Fabric administrator before connecting.

Existing higher workspace permissions are preserved. Current setup and repair
do not request a workspace role for the General Validation application.
Your browser uses your Microsoft identity to complete setup; its Fabric token
is not sent to the General Validation API.

## Connect and import tables

1. Open **Discovery**, choose **Add source**, and select **Microsoft Fabric workspace**.
2. Choose **Find my Fabric workspaces**, sign in when prompted, and select
   the workspace. Review the execution identity and requested access.
3. Choose **Register and grant Viewer access**. The first catalog refresh
   starts automatically after the connection completes.
4. Browse the Lakehouse and Warehouse tables. Tables are listed automatically
   for items with a ready SQL endpoint.
5. Select the tables you need and choose **Import selected**, or use
   **Import & prepare** for one item.

Import starts schema and table-access checks. Wait for each Dataset to become
ready before creating a Pair. Workspace visibility or a Viewer grant alone
does not establish that an imported table can run.

Owners and Contributors can refresh the catalog and import datasets. Readers
browse saved metadata without starting discovery activity.

## Recover a catalog or preparation failure

| What you see                                       | What to do                                                                                                       |
| -------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- |
| Catalog refresh failed                             | Correct the displayed access, capacity, or network issue, then use **Retry catalog refresh**                     |
| A Lakehouse or Warehouse could not list its tables | Use **List tables** on that item to retry; if polling was interrupted, use **Resume table listing** when offered |
| No SQL-visible tables                              | Check that the selected item exposes the expected tables through its SQL endpoint                                |
| An imported Dataset needs attention                | Open the Dataset and follow its schema or source-access remediation before running it                            |

A failed refresh retains the previous catalog. Tables from a failed item
listing may remain visible, but cannot be newly imported until listing
succeeds. Review the refreshed catalog before importing again.

If an older import is no longer ready, follow its preparation or re-import
instructions. Do not add unrelated storage or Azure SQL grants to resolve a
Fabric-specific error. A paused Fabric capacity or blocked network connection
needs its own correction.

## Disconnect a workspace

An Organization Owner who is also a current **Admin** of the Fabric workspace
opens the source's **Disconnect this Fabric source** section and reviews the
removal plan. If dependencies or other blockers are listed, follow the displayed
guidance and contact support if no resolution is offered. Once removal is
available, confirm the source name and choose **Remove exact grants and source**.

The flow removes matching Viewer assignments created by General Validation.
Preexisting or externally changed assignments are preserved. Earlier
product-created application assignments can also appear in the removal plan.
Removing a source does not delete its Fabric workspace or data.

If your organization used an earlier connection flow, have your Fabric
administrator review the General Validation application's remaining direct
and inherited workspace access. A connection update or registration removal
alone does not establish that earlier access has been removed.

For access or preparation errors, see [Troubleshooting](/troubleshooting-and-logs)
and [Error codes](/errors).