Error- Cant connect to the artefact

[KiaZ]

Hello,

Here is the detailed context of the issue we are facing in Microsoft Fabric right now.

What happened:

• I received an error in my semantic model indicating that a column was missing.

• I checked my dataflow gen2 and noticed that one column was not correctly written in the Delta log.

• I added the column and performed an overwrite. The dataflow then succeeded.

• Afterwards, when trying to select tables from my Lakehouse, I received artefact errors.

Actions taken:

• I deleted the problematic table.

• Re-ran the dataflow / pipeline.

• Refreshed everything (Lakehouse, SQL endpoint, browser session).

Current state:

• I even set up a new On-Premises Data Gateway.

• Despite this, I still cannot connect to any tables from Lakehouse, Excel, Exact Online, or any other source.

• Existing Lakehouse tables open, but all dataflows fail with:

  "The dataflow cannot be refreshed because of a gateway issue."

• Attempting to connect new tables hangs indefinitely at “Connection to data source.”

• Artefact issues appear when trying to open or refresh tables/dataflows.

• Report refreshes fail.

Browser refresh, SQL endpoint restart, and workspace switches do not resolve the problem.

We are looking for guidance on:

• Fully resetting connector/runtime and artefact bindings in Fabric so all tables and dataflows reconnect properly.

• Any known workarounds for Exact Online connector or artefact reference hangs.

Thank you.

[v-pnaroju-msft]

Hi KiaZ,

We are pleased to learn that your issue has been resolved. Should you have any further queries, please feel free to contact the Microsoft Fabric community.
Thank you.

[v-pnaroju-msft]

Hi KiaZ,

We would like to follow up and see whether the details we shared have resolved your problem. If you need any more assistance, please feel free to connect with the Microsoft Fabric community.

Thank you.

[KiaZ · Accepted solution]

Hi There,

The metadata in the Fabric Backend was not synced anymore, thats why i got the artefact issue. After 24h, it started to work again. So it was just a matter of waiting.

[v-pnaroju-msft]

Thankyou, @Ugk161610, for your response.

Hi KiaZ,

Based on my understanding, your column change and overwrite caused Fabric’s workspace connection or gateway binding and the affected user’s session or connection metadata to fall out of sync. This explains why one administrator can open Manage Connections while your account becomes unresponsive. Your account is using a stale or partially created connection binding that is not recognized by the backend.

Please follow the steps below, which may help to resolve the issue:

1. Sign out, clear the browser cache, or open Fabric in Incognito or Private mode and then retest Manage Connections. This forces the issuance of new tokens and often resolves per user UI freezes.
2. Ensure your account has the role of Admin, Member or Contributor in the workspace, since Viewer or Guest roles cannot bind connections. If you are a Guest, either promote your account to a tenant user or ask an internal administrator to perform the binding.
3. Attempt to delete the connection from Workspace → Settings → Connections. If the UI is blocked for your account, use the Fabric REST API to delete the stuck connection object. After deletion, reopen your dataflow or connection so that Fabric can create a fresh binding.
4. On the gateway machine, confirm that the gateway service is running, that the gateway is updated to the latest version, and that outbound traffic is permitted.
5. Open the Lakehouse SQL analytics endpoint → Manage → Refresh schema (or invoke the REST API to refresh endpoint metadata). This clears SQL endpoint caching.
6. Create a new minimal Dataflow Gen2 and attempt a refresh. If it succeeds, the workspace is functioning correctly again.

Please refer to the links below for further information:
Troubleshoot the on-premises data gateway | Microsoft Learn
Lakehouse management API - Microsoft Fabric | Microsoft Learn
Manage an on-premises data gateway | Microsoft Learn
Items - Refresh Sql Endpoint Metadata - REST API (SQLEndpoint) | Microsoft Learn
SQL Analytics Endpoint Performance Considerations - Microsoft Fabric | Microsoft Learn

We hope this information helps to resolve the issue. If you have any further queries, please feel free to contact the Microsoft Fabric community.

Thank you.

[Ugk161610]

@KiaZ ,

We’ve run into something very similar before, and the behaviour you’re describing usually happens when the Lakehouse metadata and Delta logs get out of sync with the Fabric backend. Once that happens, connectors (including Dataflows, Excel, Exact Online, etc.) start failing or hanging because they can’t resolve the internal artefact references anymore.

1. Force a rebuild of the Lakehouse metadata

This fixes most “artefact” errors.

• In the Lakehouse, go to Files > Tables

• Right-click the affected table folder in “Files”

• Rename it temporarily (for example, add _old)

• Refresh the Lakehouse view

• Recreate the table by writing from a notebook or re-running your dataflow

Fabric will rebuild the metadata and table definition from scratch.

2. Clear the corrupted Delta log state

Sometimes adding/dropping columns repeatedly leaves orphaned log entries.

In a notebook:

spark.sql("MSCK REPAIR TABLE lakehouse.tableName")

or for heavy corruption:

spark.read.format("delta").option("ignoreFileNotFound", "true")

This forces Fabric to re-index the table.

3. Regenerate SQL endpoint metadata

This step is important because the SQL endpoint keeps its own cache.

• Open the SQL endpoint

• Click Manage → Refresh schema

• Close and reopen the SQL endpoint completely

This usually clears the “can’t connect to artefact” message.

4. Delete the Lakehouse connection entry under Workspace Settings

This one is easy to miss. The connection object can get stuck pointing to an older internal ID.

• Go to Workspace settings → Connections

• Remove the connection for the Lakehouse

• Re-run your pipeline/dataflow
  Fabric will create a fresh binding.

5. For Dataflow Gen2 specifically

If every Dataflow starts failing after a schema change, it often means the runtime binding broke:

• Create a brand-new Dataflow Gen2

• Add just 1 simple source step

• Try refreshing it
  If even a brand-new one fails, then the runtime binding is definitely corrupted and a metadata rebuild (steps 1–4) is needed.

6. Exact Online connector workarounds

Exact Online is sensitive to schema changes in the Lakehouse.

GopiKrishna

[KiaZ]

@Ugk161610 

First off all, im very grateful for youre advice, i have already deleted the one ''corrupt table'' before ur reply, So does that mean that i need to apply the current spark notebook for all tables in my lakehouse? 

Ive refreshed my SQL endpoint schema and renamed my lakehouse tables but without succes... 
And how can i use the heavy corrupt notebook code?

And to run my dataflow succesfully, i need to connect my on premise gateway to my connection to run. But whenever i click on manage connections to connect it to my gateway, the screen just freezes on this page. Its the screen where u can apply ur current gateway to ur connection, without applying the gateway i cant refresh the dataflow.

But when i log in to an other admin account it doesnt, so thats weird aswell. Can u please give me some more advice on this? I can really use ur help.

Thankyou