Design a distributed message queuing system that guarantees at-least-once delivery, handles message ordering within partitions, and remains available during network partitions. Describe the replication strategy, consensus algorithm choice, and failure recovery mechanisms in detail.