Automation glossary

Retry and Backoff

A retry repeats a failed request; backoff waits progressively longer between retries so a struggling service has time to recover.

In plain language

Instead of hammering an API that just failed, the workflow waits a little, tries again, and waits longer each time.

Example

After a 429 response, a step retries after 2, 4, and 8 seconds, honoring any Retry-After header.

Rate Limit · Error Handling · Idempotency · HTTP Status Code

Common misconception

Retrying is not always safe; retrying a non-idempotent action, such as creating a payment, can create duplicates.

Implementation guidance

Retry only temporary errors such as 429 and 5xx, use exponential backoff with jitter, cap the attempts, respect Retry-After, and use idempotency keys.

Primary references

← Browse every glossary term