Track 15 · Lesson 10 of 11

Resilient automation

Make scripts survive the real world: retry transient failures with backoff, re-run safely with state files, and finish the batch when one item fails.

You will be able to

• Tell transient failures (worth retrying) from permanent ones (not worth retrying)
• Retry an operation with exponential backoff, without sleeping after the last attempt
• Make a job idempotent by recording what has already been processed
• Process a batch so one failure is reported instead of stopping everything