Worked on the rabbitmq-server repository to enhance cluster restart reliability by implementing a fail-fast mechanism in the restart process. This feature, developed using Makefile and shell scripting within build systems, ensures that if any node fails to start or come online during a cluster restart, the process is immediately aborted rather than continuing with other nodes. This approach aligns the restart logic with orchestrator expectations, reducing the risk of partial or cascading restarts and improving operational safety. The work focused on code robustness and deployment stability, delivering a targeted feature that increases predictability and observability during automated cluster management workflows.