Grant the service principal, managed identity, or organisational account at least Contributor role in the Microsoft Fabric workspace that contains the Lakehouse. Navigate to the Fabric workspace → Settings → Manage access → add the service principal with Contributor or higher. If you need the SQL endpoint to be accessible, grant Read All SQL Endpoint Data role in addition to workspace Contributor. Confirmed fix from Microsoft Learn Lakehouse connector troubleshooting documentation.

Reduce the data volume each query returns - apply explicit date range filters, select only required fields using Select Query in the advanced options rather than importing all fields, and split large Salesforce objects into smaller incremental queries. If using the Salesforce connector in a Dataflow Gen2 or semantic model with incremental refresh, ensure the incremental refresh window is narrow enough that the result set processes within 15 minutes. For very large Salesforce objects, use the Fabric Data Factory Copy Activity with the Salesforce connector (which uses Bulk API 2.0 for large extractions) rather than the Power Query/Dataflow path.

Use the Select Query advanced option in the Salesforce connector to explicitly specify only the fields required for your analysis. Remove formula fields, long text areas, and multi-select picklists that aren't needed - these consume disproportionate query capacity. In Fabric Data Factory, use $select in the Salesforce connector source settings to limit the field list. Documented fix from Microsoft Learn Salesforce connector setup documentation.

In Salesforce, navigate to Setup → Session Settings and disable Lock sessions to the IP address from which they originated. Save the change. Confirm also that API Enabled is checked in the integration user's profile settings. Note: Salesforce trial accounts do not have API access at all - a paid Salesforce licence is required. Documented fix from Microsoft Learn Salesforce connector documentation and the Microsoft connectors reference.

If you need downstream Fabric artefacts (semantic models, pipelines, reports) to be resilient to Dataverse unavailability, build a data engineering layer that copies Dataverse-sourced data from the shortcut layer into native Fabric Delta tables on a schedule. These native Delta tables persist regardless of Dataverse Link status. To re-establish a broken or inadvertently removed link: in the Power Apps maker portal, navigate to the Dataverse environment, select Link to Microsoft Fabric, and create a new Lakehouse or update the existing one. This regenerates shortcuts for all Dataverse tables in the link scope. Documented in the Microsoft Learn Dataverse Link FAQ.

In Power Apps (make.powerapps.com), select the Dataverse environment and open the Fabric Link profile. Select Refresh Fabric tables from the command bar. This triggers a table metadata refresh in the Fabric Lakehouse. The new column will appear after the next data change event occurs for that table in Dataverse - the column is added to the schema but populated on the next write event. Documented fix from Microsoft Learn Dataverse Link troubleshooting and FAQ.

Identify and remove unsupported system tables from the Dataverse Link scope via the Manage tables pane in the Fabric Link configuration. Also check for tables with logical names exceeding Fabric's supported length. Removing the unsupported table does not affect Dataverse functionality or the add-in that created it - it only removes that table from Fabric analytics sync. After removing the blocking table, the link will be able to recreate shortcuts and generate metadata tables for the remaining scoped tables normally. Going forward, review new solution components (from add-ins, ISV packages, and Microsoft updates) before adding them to the Fabric link scope.

Confirm the Azure region of your Dataverse environment in the Power Platform admin centre. Provision a Fabric capacity (F SKU) or Power BI Premium capacity (P SKU) in the same Azure region. Power BI Premium Per User is not sufficient for Dataverse Link. A Fabric trial capacity can be used for testing. If the capacity exists in a different region, the Dataverse environment must be migrated to match, or a new Fabric capacity must be provisioned in the correct region.

Add explicit partition and filtering to the SAP Table connector source settings to reduce the data volume per extraction run. Use $top row limits and partition by date range, company code, or plant to break the extraction into smaller batches that fit within SAP memory limits. For very large SAP tables, consider switching from the SAP Table connector (which uses RFC) to the SAP CDC connector (which uses change data capture at the database level and does not load entire tables into application server memory). The SAP HANA connector's direct database connection also avoids this RFC memory constraint. Check with your SAP Basis team on the application server memory configuration if partitioning alone doesn't resolve it.

Confirm your Zoho CRM data centre: log in to Zoho CRM, navigate to Setup → Company Details → Data Centre. Match the API base URL and OAuth accounts URL to that data centre. If the data centre is EU, use accounts.zoho.eu for the OAuth token endpoint and https://www.zohoapis.eu/crm/v8/ for all API calls. Update both the linked service in Fabric Data Factory and any Key Vault secrets storing the API domain to use the correct regional URL. Regenerate the OAuth tokens using the correct regional accounts URL if existing tokens were generated against the wrong domain.

Verify that the source schema contains the target column for the multi-target lookup field. In the Fabric pipeline column mapping for the Dynamics connector, add the target column explicitly in the format {fieldName}@EntityReference. This tells the connector which entity type to resolve the lookup to. If you don't need the entity reference in the Fabric output, exclude the multi-target lookup field from the column mapping entirely. Documented fix from Microsoft Learn Dynamics/Dataverse connector troubleshooting.

Manually add the missing columns in the column mapping tab of the Fabric pipeline activity. Do not rely solely on the auto-imported schema for Dynamics entities - always cross-reference the entity's full field list in Dynamics 365 or Dataverse and add any columns that aren't automatically detected. Ensure added columns are in the correct format (data type and column name matching the Dynamics API field name). Documented fix from Microsoft Learn Dynamics connector troubleshooting.

Enable schema evolution on the target Delta table by setting the mergeSchema option in the Fabric pipeline sink settings or notebook write operation. In a Fabric notebook: df.write.option("mergeSchema", "true").mode("append").format("delta").save(path). In a Fabric pipeline Copy Activity: enable Schema evolution in the sink settings. For type mismatches (not just new columns), mergeSchema alone won't resolve it - cast the incoming data to match the existing table schema in a transformation step before writing. Documented in Microsoft Learn Lakehouse troubleshooting.

For Power BI Desktop (January 2026+): update to the latest Power BI Desktop release - encryption settings were adjusted in subsequent releases to handle Fabric endpoints correctly. If connecting via a custom application or driver: update to the latest Microsoft OLE DB Driver for SQL Server or ODBC Driver 18 for SQL Server, which support the encrypted connection requirement. For the Fabric Lakehouse SQL endpoint specifically, ensure the connection string includes Encrypt=True;TrustServerCertificate=False and that the client machine trusts the certificate chain for the Fabric endpoint domain. Documented in the Fabric community (March 2026).