Horizontal Pod Autoscaler - CI/CD Glossary Definition

A Horizontal Pod Autoscaler (HPA) is a Kubernetes controller that automatically changes the number of pod replicas in a workload based on observed metrics such as CPU utilization, memory, or custom metrics.

How it decides

The HPA periodically compares a target metric (for example 70% CPU) against the current average across pods and scales the replica count up or down to converge on the target, within configured min/max bounds.

Scaling CI runners

For self-hosted runners on Kubernetes, scaling is usually driven by queue depth (pending jobs) rather than CPU. Tools like the Actions Runner Controller or KEDA add a runner pod per queued job, which is closer to event-driven autoscaling than a CPU-based HPA.

Related guides