What Happens to Bad Records?

Dead Letter Queues and Recovery This question tests your failure recovery instincts. The trap: 'we skip bad records and keep going.' The red flag: 'the pipeline fails and we fix it manually.' The senior answer: 'Bad records go to a dead letter queue. The pipeline continues processing good records, the bad records are preserved for investigation, and we can reprocess them after fixing the root cause.' A dead letter queue (DLQ) gives you three things interviewers care about: the pipeline keeps running (no full failure for one bad record), you don't lose data (it's in the DLQ), and you have an audit trail for investigation. When the interviewer asks 'what happens to the bad records,' say those three things in that order. The follow-up will be: 'How do you backfill after a failure?' This is wh