Kubernetes Cluster Autoscaler Support (Experimental)

By default, Longhorn blocks Kubernetes Cluster Autoscaler from scaling down nodes because:

• Longhorn creates PodDisruptionBudgets for all engine and replica instance-manager pods.
• Longhorn instance manager pods have strict PodDisruptionBudgets.
• Longhorn instance manager pods are not backed by a Kubernetes built-in workload controller .
• Longhorn pods are using local storage volume mounts.

For more information, see What types of pods can prevent CA from removing a node?

If you want to unblock the Kubernetes Cluster Autoscaler scaling, you can set the setting Kubernetes Cluster Autoscaler Enabled.

When this setting is enabled, Longhorn will retain the least instance-manager PodDisruptionBudget as possible. Each volume will have at least one replica under the protection of an instance-manager PodDisruptionBudget while no redundant PodDisruptionBudget blocking the Cluster Autoscaler from from scaling down.

When this setting is enabled, Longhorn will also add cluster-autoscaler.kubernetes.io/safe-to-evict annotation to Longhorn workloads that are not backed by a Kubernetes built-in workload controller or are using local storage mounts.

Warning: Replica rebuilding could be expensive because nodes with reusable replicas could get removed by the Kubernetes Cluster Autoscaler.

© 2019-2024 Longhorn Authors | Documentation Distributed under CC-BY-4.0