RELEASED: Public: October 18, 2023
New features
- When running a controller in High Availability (HA) mode, provision a pod disruption budget (BEE-29191)
-
Using a pod disruption budget ensures that at least one replica is always up when scaling down node pools.
- Omit
cluster-autoscaler.kubernetes.io/safe-to-evict=falseon controllers in High Availability (HA) mode (BEE-29423) -
When using High Availability (HA) mode, you can evict replicas for the purpose of scaling down a node pool as it does not cause loss of availability.