What happens to workloads in Kubernetes that exceed defined resource limits?

When workloads in Kubernetes exceed their defined resource limits, the correct outcome is that these workloads are automatically terminated. This behavior is part of Kubernetes' resource management system, which is designed to maintain cluster stability and efficiency.

Resource limits in Kubernetes are set to prevent any single pod from consuming too many resources—such as CPU or memory—which could negatively impact the performance of other pods running on the same node. When a pod exceeds its allocated resources, the Kubernetes scheduler identifies this resource over-consumption and will terminate the pod. This action serves to maintain the integrity of the overall cluster and allows Kubernetes to enforce its resource quotas effectively.

In instances where the workload is critical, Kubernetes can attempt to restart the terminated pod automatically, depending on the configuration of the pod’s restart policy. This ensures that the application can recover from transient issues while still adhering to the defined resource constraints.

The other potential scenarios referenced by the remaining options do not accurately represent Kubernetes' behavior regarding resource limits. For example, workloads being paused or migrated is not a standard behavior in response to exceeded limits, and continuing to run despite exceeding those limits contradicts the resource management principles that are fundamental to Kubernetes.