Glossary · Data, timing and communication
Retry
Also known as: retry attempt
In systems engineering, a retry is a repeated attempt after a failed or missing communication or operation, usually bounded by a maximum number of attempts.
- Data and interfaces
- Integrated systems
In one sentence
A retry repeats a failed or unanswered request; safe retries need idempotent operations, a backoff rule and a defined end.
Example
A label printer on a packaging line does not acknowledge a print job; the line controller resends it twice and then stops the labeler and raises an alarm.
How it applies
- Integrated systems: Specify when to retry (after which errors), how often, with which backoff, and what happens after the last attempt. Only retry operations that are idempotent or protected against duplicates.
- Functional safety: Retries hide faults and add delay. Count the worst-case retry time in response-time budgets, and make sure repeated failures lead to a fault reaction rather than endless retrying.
- Evidence: Log retries. A rising retry rate is often the first sign of a degrading network or device.
Retry vs. timeout
A timeout decides that an attempt has failed; a retry is the next attempt. The number of attempts, the timeout and the backoff together determine how long a failure can remain unhandled.