Worked on Apache RocketMQ, delivering a targeted fix to improve RocksDB high-availability synchronization between master and slave brokers. Addressed a configuration consistency issue by ensuring that topic and subscription group changes, including deletions and updates, propagate correctly from master to slave, thereby reducing configuration drift and enhancing cluster stability during failover scenarios. The solution involved backend development in Java, with a focus on distributed systems and configuration management. By refining the synchronization logic within the repository, the work improved data replication reliability and cluster consistency, contributing to more robust high-availability operations without introducing new features during the development period.