Pushstuck

Pushstuck refers to a condition in push-based or asynchronous messaging systems where producers continue to push data but the downstream path cannot progress, resulting in a backlog that prevents new pushes from advancing. In such scenarios, broker queues, stream partitions, or consumer pools fall behind the rate of input, causing increased latency and resource pressure across the system.

The term is informal and primarily appears in software engineering discussions, issue trackers, and forums. It is not an official standard but describes a recognizable failure mode in modern event-driven architectures.

Causes include mismatched throughput between producers and consumers, insufficient backpressure or flow control, slow or unavailable downstream services, network bottlenecks, large message sizes, bursty traffic, partition skew, and misconfigured batching, retry, or retry backoff policies.

Effects can include rising queue depths, longer end-to-end latency, higher memory consumption due to buffered data, increased pressure on storage and processing services, and systemic degradation. If buffers overflow, systems may trigger throttling or data loss unless safeguards such as dead-letter queues are in place.

Detection relies on monitoring metrics like queue depth, consumer lag, processing rate, error rates, and tail latency. Mitigation strategies include applying backpressure to producers, rate limiting, auto-scaling consumers, optimizing batch sizes, implementing idempotent processing, using durable queues, and deploying dead-letter queues to isolate problematic data sources.

See also: backpressure, flow control, message queue, dead-letter queue, backoff strategy.