Worked on stabilizing the restore workflow in the apache/doris repository, focusing on backend development and database management using Java. Addressed a critical bug by introducing a staging mechanism for restored tables, which prevents premature deletion and correctly manages aliased table names during disaster recovery operations. Enhanced metadata synchronization between master and follower nodes, ensuring data integrity and reducing the risk of metadata divergence. The solution clarified restore steps and improved error visibility, strengthening operational safeguards for cluster stability. This work emphasized careful bug fixing and robust process design, contributing to more reliable disaster recovery and consistent metadata management across distributed systems.