🔥 3-day streak
Professional Data Engineer147 / 169
Question 147 of 169
A streaming Dataflow pipeline reads from a Pub/Sub subscription and writes aggregated metrics to BigQuery. After a production deployment, the team discovers that a transformation bug produced incorrect results for the last 6 hours. The subscription retains acknowledged messages, and message retention is set to 7 days. After deploying the corrected pipeline code, the team wants to reprocess the last 6 hours of messages without asking upstream publishers to resend anything. What is the most appropriate approach?
Reviewed for accuracy · Report an issueNext question