Databricks introduces Session Restore for serverless jobs in Beta
Databricks has launched Session Restore for serverless compute jobs, currently available in Beta. This feature enables developers to restore Python variables and Spark sessions from failed, long-running, or cancelled job runs directly into a new interactive notebook. It significantly aids in debugging failures and exploring results without needing to re-execute the entire job. Databricks automatically captures state snapshots under specific conditions and retains them for up to seven days.
Features (1) ›
- Session Restore for Databricks Serverless Jobs
This feature allows users to restore Python variables and Spark sessions from serverless compute job runs into a new interactive notebook. This capability helps debug failures or explore results without rerunning the entire job. Databricks automatically captures a state snapshot when an eligible run fails, exceeds 30 minutes, or is cancelled, keeping it for 7 days.
https://docs.databricks.com/aws/en/release-notes/product/2026/august#session-restore-for-serverless-jobs-is-now-in-beta
Related releases
- Databricks Connector for Google Sheets Adds Write-Back to Unity Catalog Databricks Release Notes ·
- Databricks enables base environments for Python dependency management on classic compute (Beta) Databricks Release Notes ·
- Databricks to deprecate root_cause_analysis column in data quality monitoring results Databricks Release Notes ·
- Databricks Genie One Enhancements: Consistent Scheduled Tasks, PDF Export, and Mobile App Improvements Databricks Release Notes ·
- Databricks Lakeflow Connect Adds Workiva Connector (Beta) Databricks Release Notes ·
- Databricks RBAC to Default for Workspaces with Compliance Security Profile Databricks Release Notes ·