The original title is "How We Make Every Scheduled Job Safe to Re-Run" which is 9 words. Let me check the rules:

Short summary

Scheduled jobs often re-run due to deploys, overlaps, or server bounces, causing duplicate writes and actions. This article presents two idempotency patterns: dedup keys with MongoDB upsert + $setOnInsert for atomic find-or-create, and a ledger collection that tracks used items before processing to prevent generators from handing out duplicates on re-run.

• •Dedup keys with MongoDB upsert + $setOnInsert for atomic find-or-create
• •Ledger collection tracks used items before processing to prevent duplicate generation
• •Ledger writes before processing so crashes skip rather than duplicate

Generated with AI, which can make mistakes.