cloudflare-queues

Background jobs with Cloudflare Queues use a producer-consumer pattern: create a queue, deploy a producer to send messages from your Workers, then deploy a consumer to process them. Messages are reliable and support retries with exponential backoff and dead-letter queues for failed items.

Dead-letter queues capture messages that fail after retries. Configure a DLQ alongside your main queue, set retry policies with exponential backoff, and monitor the DLQ to diagnose processing failures and prevent data loss.

Yes, batch processing is built in. Set max_batch_size and max_batch_timeout to group messages and process them together, reducing overhead and improving throughput for workloads like analytics and order processing.

Individual messages are limited to 128 KB. For larger payloads, store the data in R2 and pass a reference through the queue, allowing you to process large files without hitting message size constraints.

Cloudflare Queues supports both explicit and implicit acknowledgement modes. Explicit mode gives you control over when messages are marked complete; implicit mode automatically acknowledges after successful consumer execution.

Queues excel at email sending, analytics aggregation, order processing, and data pipelines—any workload requiring reliable background execution, batching, or retry logic without blocking user responses.