Worked on improving topology stability and reliability for the slackhq/vitess repository, focusing on ZooKeeper connection management during disconnect scenarios. Addressed a bug where ZooKeeper topology connections were not properly closed after disconnection events, such as server restarts or state transitions. Implemented connection cleanup logic in Go, ensuring that old connections are reliably terminated to prevent resource leaks. Added regression tests to validate that connections are closed as expected, strengthening the topology lifecycle under various states like Disconnected, Expired, and Connecting. These changes enhanced cluster resilience and reliability, particularly during failover and restart scenarios, using skills in Go, ZooKeeper, and testing.