When did you detect a mismatch?

I was working on a nightly ingestion pipeline that moved customer transaction data from a legacy database into a data lake. During the validation step, I compared the row counts and a checksum between the source and destination. The counts matched, but the checksum differed, indicating a data corruption issue. I logged the discrepancy, paused the pipeline, and investigated the source system logs. It turned out a network timeout had caused a partial write, leaving the source table in an inconsistent state. I corrected the source data, re‑ran the ingestion job, and added a checksum verification step to the pipeline to catch similar issues early. This experience taught me the importance of having multiple validation layers and the ability to quickly diagnose and remediate data mismatches.