Getting Started Guides Architecture API Changelog

Choose SDK

Java/Guides/

Reliability

Harden your Java task queue for production workloads.

← PreviousTopic Pub/Sub Next →Error Handling

Error HandlingTask failure lifecycle, error inspection, the exception hierarchy Delivery GuaranteesAt-least-once delivery and exactly-once patterns RetriesAutomatic retries with exponential backoff TimeoutsBound task runtime per execution attempt IdempotencyDeduplicate enqueues with unique keys DebouncingCollapse a burst of enqueues into one run on a sliding deadline Flow controlDebounce vs throttle vs coalesce, and what to do with excess load Rate LimitingThrottle and defer enqueues with gates ConcurrencySize and autoscale the worker's handler pool Dead-letter queueCapture and replay exhausted jobs Distributed LockingMutual exclusion across workers with database-backed locks