Handling Duplicate Background Jobs in .NET with Retry-Safe Logic

Most durable job queues guarantee at-least-once delivery. That is the right default — losing a job is worse than running it twice. But it means your handler will, eventually, be invoked twice for the same logical work.

Wrap each unit of work in an envelope that carries a stable id derived from the business event, not the queue message. On entry, the handler checks whether that id has already been processed and short-circuits if so.

This single pattern absorbs most duplicate-delivery cases without making individual handlers harder to read.

When the downstream system is not idempotent, isolate the call behind a small adapter and add your own dedupe at that boundary. Log the correlation id you used so support can trace what happened.