diff --git a/Documentation/Upgrade/rook-upgrade.md b/Documentation/Upgrade/rook-upgrade.md index a97286a3f..160ff819f 100644 --- a/Documentation/Upgrade/rook-upgrade.md +++ b/Documentation/Upgrade/rook-upgrade.md @@ -52,34 +52,20 @@ those releases. official releases. Builds from the master branch can have functionality changed or removed at any time without compatibility support and without prior notice. -## Breaking changes in v1.18 - -* Helm: Two new properties have been added to the storage classes in the `rook-ceph-cluster` chart. - Adding these properties will cause the helm upgrade to fail since storage classes are immutable. - Before upgrading the `rook-ceph-cluster` chart, one of these actions is necessary: - - Delete the existing StorageClasses. The upgrade will create the storage classes with the new properties. - Existing PVs will not be affected by the storage class temporarily being deleted. - - Remove the newly added parameters (`controller-publish-secret-name` and `controller-publish-secret-namespace`) - from your StorageClasses prior to upgrading. The new parameters are only needed if fencing is enabled. - - For more background, see [PR #16442](https://github.com/rook/rook/pull/16442). +## Breaking changes in v1.19 * The minimum supported Kubernetes version is v1.29. -* The **CSI operator** is required for RBD, CephFS, and NFS volumes. Previously, the CSI driver was automatically configured - by Rook. Now the CSI operator has been factored out of Rook to run independently to manage the Ceph-CSI - driver. During the upgrade and throughout the v1.18.x releases, Rook will automatically convert any Rook CSI settings - to the new CSI operator CRs. This transition is expected to be completely transparent. - Starting in v1.19, the Rook settings related to CSI will be deprecated and there will be a guide on how to configure the new CSI operator CRs directly. - * The CSI operator is created directly via `csi-operator.yaml`, or by the helm setting `csi.rookUseCsiOperator` - in the `rook-ceph` chart. - * The new CSI operator CRs are `cephconnections.csi.ceph.io`, `drivers.csi.ceph.io`, - `operatorconfigs.csi.ceph.io`, `clientprofiles.csi.ceph.io`, and `clientprofilemappings.csi.ceph.io`. - * The CSI operator has been in experimental mode since Rook v1.15 and we have confidence in its stability. That said, if - there is any blocking issue with the csi operator, the previous CSI driver can be re-enabled by setting - `ROOK_USE_CSI_OPERATOR: false` in operator.yaml or applying the helm setting `csi.rookUseCsiOperator: false`. +* Helm: The rook-ceph-cluster chart has changed where the Ceph image is defined, to allow separate settings + for the repository and tag. If you deploy with a custom values.yaml: + 1. Remove the [cephVersion](https://github.com/rook/rook/blob/release-1.18/deploy/charts/rook-ceph-cluster/values.yaml#L102-L114) section from the `cephClusterSpec`, to avoid a conflict with the new settings. + 2. If desired to override the Ceph image repository or tag, apply the new [cephImage](https://github.com/rook/rook/blob/master/deploy/charts/rook-ceph-cluster/values.yaml#L88-L102) settings such as: -* Rook now validates node topology during CephCluster creation to prevent misconfigured CRUSH hierarchies. For example, if [topology labels](../CRDs/Cluster/ceph-cluster-crd.md#osd-topology) like `topology.rook.io/rack` are duplicated across zones, cluster creation will fail. The check applies only to new clusters. Existing clusters will only log a warning in the operator log and continue. +```yaml +cephImage: + repository: quay.io/ceph/ceph + tag: v19.2.3 +``` ## Considerations