The assertion to ensure strict timestamp usage fired in oplog replay after rollback to stable. The symptom is that we have an out of order timestamp with prev_durable_ts larger than start_ts. The next update of it is an abort update without any timestamp. We assume the prev_durable_ts comes from the onpage value. However, we cannot confirm that in the core dump. Also we looked at the disk files after rollback to stable. There is no data with timestamp larger than the stable timestamp and update with prev_durable_ts. Therefore, the data with prev_durable_ts seems to be written after rollback to stable and rollback to stable is working as expected.

Scope: Investigate where the prev_durable_ts comes from and is there any issue with WiredTiger. If its a WT issue, fix it or reassign the BF to the respective team.