I fixed the issue, the problem was I was only running 2 rabbitmq nodes. This caused entire cluster to go down while one node is down. When I increased the node count to 3 the rolling update worked.