New: Cookbooks and AI ExplanationsStep-by-Step recipes to solve problems connected to Roadmaps and Cheat Sheets. Need more details? Use AI buttons for structured and simple explanations with concrete examples throughout the whole platform.Take a look
A payment API has one bad ten minutes, and a thousand jobs that all failed in the same second all retry in the same second.
What you'll have at the end
A job with capped exponential backoff and random jitter, tested against a stub that fails the first three calls on purpose
You need
A task that already runs behind a queue and calls an outside API as part of its work, with a retry in place that waits the same flat amount every time and has no upper limit on how large that wait can grow.
Not covered
What a job does once it has used up every retry and still hasn't succeeded; that failure needs a queue of its own to land in, a separate problem from how long each attempt waits.
Leans on
A missing dead letter queue means an exhausted job just vanishes
covers what happens once a job has used every retry this recipe grows and still hasn't succeeded, since a longer wait never turns a permanent failure into a temporary one.
Catch the one poison message that retries forever and starves every worker
covers a message that never succeeds no matter how long it waits, unlike a temporary outage that eventually clears.
Checked 19 Aug 2026
Part of the Background Job Processing Systems cookbook