kube-amqp-autoscale

kube-amqp-autoscale adjusts the number of workers according to how many messages are waiting in an AMQP queue. It is useful for applications where queued work is a better measure of demand than processor utilization. The controller samples the queue and applies configured thresholds and adjustment limits to a Kubernetes workload. Its historical alpha implementation has specific scaling behavior, so a replacement needs to match those assumptions rather than copy the same target number blindly.

Current guidance

The recovered description matches the mbogus/kube-amqp-autoscale README, including its warning to use HPA for CPU-bound workloads. The historical upstream field pointed to Kubernetes’ HPA documentation, which is a background reference rather than this implementation’s source. The matching project explicitly labels its status alpha.

Its configuration uses AMQP queue statistics, averaging intervals, a threshold and per-iteration increase or decrease limits to scale a named Kubernetes workload. Identify those semantics before replacing it with a metrics-driven HPA or KEDA scaler; equal-looking queue targets can produce different replica decisions and activation behavior.

Check broker credentials, virtual hosts and TLS, and verify that only the intended Deployment or other scale target can be changed. Test missing statistics, an empty queue and a backlog containing slow messages. Worker acknowledgement, retry and idempotency behavior still determines whether processing is reliable. This review does not establish a current release/support matrix for the alpha controller, and the README’s insecure or legacy API options should not be copied into a new cluster without a security and compatibility review.