An Apache Spark-based analytics platform optimized for Azure.
Open an Azure support request immediately for backend investigation, but plan for rebuilding the workspace because Azure Databricks documents deletion as irreversible even during its seven-day soft-deleted state.^1^
Request the only possible investigation path
Because the deletion occurred today, preserve the current resources and submit the case now:
- In the Azure portal, select ? → Help + support → Create a support request.
- Select the affected subscription and Azure Databricks as the service.
- Choose the Recover a deleted workspace problem category if offered.
- State clearly:
- Workspace name and full Azure resource ID.
- Workspace ID.
- Original resource group, subscription, and Azure region.
- Exact deletion time in UTC.
- Whether deletion occurred by deleting the workspace, deleting its resource group, or using a force-deletion option.
- That the resource group and storage account have already been restored.
- That the requested outcome is backend assessment of any limited workspace recovery, not storage-account recovery.
- Include the deletion operation from Azure Monitor → Activity log, if available.
- Select the highest severity justified by the production impact and permit advanced diagnostics if organizational policy allows.
Creating a subscription-level technical request requires Owner, Contributor, Support Request Contributor, or equivalent Microsoft.Support/* permissions, and technical investigation requires an eligible support plan.^2^
Do not delete, rename, redeploy over, or modify the remaining managed resources while support assesses the case.
What recovery can realistically provide
The documented product behavior is:
- Associated compute resources, disks, and networking configurations are cleaned up when the workspace is deleted.
- The workspace remains soft-deleted for seven days, but that state does not provide a customer-accessible restore operation.
- Microsoft explicitly states that a workspace cannot be recovered or restored after entering either soft-deleted or hard-deleted state.^1^
There has been an exceptional Microsoft Q&A case where the backend team restored a deleted workspace with limited resources, but the workspace could not be fully returned to its previous configuration.^3^ This is not a documented recovery guarantee. Support must determine whether anything remains recoverable in this specific deletion.
Therefore, assume the following unless support confirms otherwise:
| Asset | Expected recovery path |
|---|---|
| Notebooks and workspace files | Only from Git, prior exports/backups, or an exceptional backend recovery |
| Jobs and cluster configurations | Recreate from Terraform, deployment code, exported JSON, or documentation |
| Workspace settings, ACLs, secret scopes | Recreate; secret values should come from the original secret system, such as Key Vault |
| VMs, disks, and cluster runtime state | Not recoverable |
| Data in customer-owned ADLS containers | Usually remains independently accessible, subject to its own deletion and retention state |
| Retained Unity Catalog data files | Potentially reusable, but catalog metadata and permissions must be assessed separately |
What the restored storage account can and cannot provide
A restored storage account does not by itself restore the Databricks control-plane record represented by the old workspace ID. Azure Databricks separates:
- Workspace file-system assets: notebooks, SQL queries, dashboards, repos, libraries, and small workspace files.
- Workspace system data: job results, notebook revisions, cluster logs, and internal metadata.
- Customer data objects: such as Unity Catalog tables, volumes, external tables, and files.
For classic workspaces, the workspace also depends on Databricks control-plane databases; workspace system data is distinct from DBFS even when both use the same storage account.^4^
Consequently:
- Preserve the restored account unchanged. Take an inventory of containers and paths before attempting reuse.
- Determine whether
ade28adls11was:- Customer-owned ADLS connected to Databricks;
- The Databricks workspace storage account; or
- Both through different containers.
- Copy recoverable customer files to a separate protected location before attaching them to a replacement workspace.
- Do not assume unfamiliar UUID-based or internal paths are notebooks or job definitions. Their presence does not reconstruct the missing control-plane metadata.
- Ask support specifically whether the retained storage belongs to the old workspace and whether attaching or modifying it would interfere with backend recovery.
If the default Unity Catalog workspace catalog was enabled and deletion was not forced, its storage container and access connector can be retained after workspace deletion; that preserves associated data files, not necessarily the deleted workspace’s notebooks, jobs, settings, or metadata.^1^
If backend recovery is unavailable
Create a new workspace and rebuild from Git repositories, Terraform/Bicep, CI/CD artifacts, exported workspace files, audit records, and the preserved customer-owned data. For future protection, export notebooks, jobs, clusters, ACLs, and other configuration before decommissioning and keep those exports outside the workspace.^5^
References