Unable to delete ghost dabatricks workspace

Hello Azure Administrator,

Thanks for raising this question in Azure Q&A forum.

A "ghost" Azure Databricks workspace that appears in the Azure portal but has no URL, no managed resource group, and cannot be deleted is a known provisioning failure state where the workspace resource remains in a "stuck Accepted" provisioning state indefinitely. This is not recoverable through standard CLI/Portal deletion and requires Microsoft engineering intervention.

Why this happens

The workspace got stuck during the Appliance deployment phase (the internal Databricks control plane VM deployment). The Azure Databricks resource provider failed to complete the workspace URL assignment and managed resource group creation, leaving the resource in a limbo state where normal deletion operations fail with errors about "resource still provisioning" or "deny assignment from managed RG" (even though no managed RG actually exists).

Step 1 — Confirm the stuck state

Verify the exact symptoms:​

bash
az databricks workspace show --name dabatricks-workspace --resource-group <rg> --query "{provisioningState:provisioningState, workspaceUrl:workspaceUrl}"

Expected output:

text
{
  "provisioningState": "Accepted",
  "workspaceUrl": null
}

Step 2 — Attempt force deletion (will likely fail but required)

bash
# CLI force delete

These will fail with errors about provisioning state or deny assignments.

Step 3 — Open an Azure Support Ticket (the only fix)

This is a platform-level issue that requires Azure Databricks resource provider engineering to purge the stuck appliance deployment from the backend. Create a Severity A support ticket immediately:

Support Ticket Details to Include:

text
Title: Azure Databricks workspace stuck in "Accepted" provisioning state - cannot delete

Symptoms:
1. az databricks workspace show shows provisioningState: "Accepted", workspaceUrl: null
2. No workspace URL accessible, no managed resource group created
3. Delete operations fail with [exact error message]
4. No active clusters/jobs running

Request:
Escalate to Azure Databricks resource provider engineering team to:
- Purge the stuck appliance provisioning state
- Remove any backend deny assignments preventing deletion
- Allow clean deletion of the ghost workspace resource

Resource Details:
- Workspace: dabatricks-workspace
- RG: <your-rg>
- Subscription: <sub-id>
- Region: <region>

Step 4 — Post-ticket cleanup (after engineering unblocks)

Once Microsoft clears the backend state:

bash
# Delete the workspace resource

Why standard methods don't work

Azure Portal delete waits for provisioningState to change from "Accepted" — which never happens

CLI/PowerShell force delete fails due to backend deny assignment from the incomplete appliance deployment

Resource Group deletion fails due to the workspace resource dependency

Prevention

Always monitor workspace provisioning with az databricks workspace show during initial deployment. If it stays "Accepted" for >2 hours with workspaceUrl: null, cancel immediately before it becomes a ghost resource.​

If it helps kindly accept the answer.

Best Regards,

Jerald Felix