Replay's implementation faces challenges due to its reliance on partial determinism, which complicates the debugging of crash-fixes by causing divergences between recorded and replayed sessions. The development process heavily utilizes Honeycomb for tracking and analyzing system crashes, with a triage script categorizing recorded crash data into specific issues. This method helps in identifying patterns and managing uncategorized mismatches, which occur when expected entries in the recording stream do not align with the replay process. The article discusses the intricate process of debugging these mismatches, such as a crash caused by discrepancies in lock acquisitions between the recording and replay. Despite the challenges, measures like recording assertions are employed to improve crash tracking and analysis. The author acknowledges the ad-hoc nature of the current debugging approach and expresses a desire for more robust infrastructure. They invite individuals interested in complex systems work to join their team, emphasizing the need for innovative solutions to enhance Replay's robustness for widespread use.