This article explains how the Ocean Autoscaler calculates and works with Kubernetes workloads that request ephemeral storage. It describes the considerations related to ephemeral-storage requests and scaling behavior in an Ocean integrated cluster.

For utilizing ephemeral storage in Ocean clusters, the available ephemeral storage is determined by the root volume size of the cluster. During scale-up, 20% of the available ephemeral storage is reserved and deducted from the total capacity.

Pods deployed in an Ocean integrated cluster must have their ephemeral-storage requests fall within the range of the total ephemeral storage minus this 20% reduction. If a pod's request exceeds this threshold, the pod will remain in a pending state.