6 terms
Checkpointing saves the progress and state of an automated workflow or job at defined points, so it can resume after a failure instead of starting over.
A dead letter queue holds messages that a system could not process or deliver after its retry limits, isolating them for inspection and reprocessing.
Exponential backoff is a retry strategy that multiplies the wait after each failed attempt, usually with jitter and a cap, to protect overloaded services.
Graceful degradation is a reliability design in which an automated system keeps its core functions running with reduced capability when a dependency fails.
Idempotency is the property of an operation that produces the same result whether it runs once or many times, which makes retries in automation safe.
A retry policy defines when, how often and how quickly an automated system repeats a failed operation before giving up or escalating the error.