v2.12.2
Note: If using the recently added HostPath volume mounts, this version will cause a cluster restart as internal volume names have changed
Changed
- Updated helm chart generation scripts to use new
helmoverlay for better templating. - Operator now correctly detects endpoint changes in AIS clusters and recreates the client when the URL (including ports) changes.
- Removed limitations on path length for HostPath volume mounts by using a constant, indexed volume name for HostPath volumes within pods.
- Fixed a bug where if all cluster nodes and state are lost but the CR is persisted in K8s, clusterID in status would not update to new cluster.
- Reduced scope of ClusterRole
manager-role - Tightened default securityContext for manager pod and container
Added
spec.targetSpec.pdboption to configure a PodDisruptionBudget for target pods.
Full changelog v2.11.0...v2.12.2