将ETL管线生产化：三次失败揭示数据工程不仅是写脚本
英文摘要
The author attempted to make an ETL pipeline production-ready. Three things broke during the process. Each failure taught insights that scripting alone could not provide. The experience highlighted the additional complexities of production data engineering.
中文摘要
作者试图将一条ETL管线推向生产，过程中三处环节失败。每一次失败都提供了单纯写脚本无法获得的教训。此次经历凸显了生产环境中数据工程额外的复杂性。
关键要点
Making an ETL pipeline production-ready reveals failures unseen in scripting.
将ETL管线推向生产会暴露脚本阶段未见的故障。
Each failure provides specific lessons about reliability, recovery, or edge cases.
每次失败都提供了关于可靠性、恢复或边缘情况的具体教训。
Scripting alone does not prepare one for production data engineering challenges.
仅凭脚本编写经验不足以应对生产数据工程的挑战。