Redpanda, a new streaming storage engine, underwent rigorous testing with Jepsen to ensure its reliability and consistency in distributed systems. Redpanda employs verified protocols like Raft and two-phase commit to safeguard data despite failures, but theoretical soundness doesn't always guarantee flawless implementation. Partnering with Jepsen, a company known for auditing distributed systems, Redpanda aimed to validate its consistency and availability. The Jepsen tests, which lasted several months, revealed that while Redpanda's consensus layer is robust, there were issues in the idempotency and transactional layers that have since been addressed. The remaining consistency challenges relate to inherent properties of the Apache Kafka protocol rather than implementation flaws. Additionally, Jepsen identified some availability issues that Redpanda is investigating, though they don't result in data loss or event reordering. This partnership allowed Redpanda to enhance its documentation, improve in-house chaos testing, and gain insights into the differences between Kafka and database transactional models, ultimately contributing to a safer and more efficient platform.