Rook can be installed on any existing Kubernetes cluster as long as it meets the minimum version and Rook is granted the required privileges (see below for more information).
Kubernetes v1.21 or higher is supported.
Architectures supported are
amd64 / x86_64 and
To configure the Ceph storage cluster, at least one of these local storage types is required:
- Raw devices (no partitions or formatted filesystems)
- Raw partitions (no formatted filesystem)
- LVM Logical Volumes (no formatted filesystem)
- Persistent Volumes available from a storage class in
Confirm whether the partitions or devices are formatted with filesystems with the following command:
FSTYPE field is not empty, there is a filesystem on top of the corresponding device. In this example,
vdb is available to Rook, while
vda and its partitions have a filesystem and are not available.
Enabling the Rook admission controller is recommended to provide an additional level of validation that Rook is configured correctly with the custom resource (CR) settings. An admission controller intercepts requests to the Kubernetes API server prior to persistence of the object, but after the request is authenticated and authorized.
To deploy the Rook admission controllers, install the cert manager before Rook is installed:
Ceph OSDs have a dependency on LVM in the following scenarios:
- If encryption is enabled (
encryptedDevice: "true"in the cluster CR)
metadatadevice is specified
LVM is not required for OSDs in these scenarios:
- OSDs are created on raw devices or partitions
- OSDs are created on PVCs using the
If LVM is required, LVM needs to be available on the hosts where OSDs will be running. Some Linux distributions do not ship with the
lvm2 package. This package is required on all storage nodes in the k8s cluster to run Ceph OSDs. Without this package even though Rook will be able to successfully create the Ceph OSDs, when a node is rebooted the OSD pods running on the restarted node will fail to start. Please install LVM using your Linux distribution's package manager. For example:
- Since version 1.5.0 LVM is supported
- Logical volumes will not be activated during the boot process. You need to add an runcmd command for that.
Ceph requires a Linux kernel built with the RBD module. Many Linux distributions have this module, but not all. For example, the GKE Container-Optimised OS (COS) does not have RBD.
Test your Kubernetes nodes by running
modprobe rbd. If the rbd module is 'not found', rebuild the kernel to include the
rbd module, install a newer kernel, or choose a different Linux distribution.
Rook's default RBD configuration specifies only the
layering feature, for broad compatibility with older kernels. If your Kubernetes nodes run a 5.4 or later kernel, additional feature flags can be enabled in the storage class. The
object-map features are especially useful.
If creating RWX volumes from a Ceph shared file system (CephFS), the recommended minimum kernel version is 4.17. If the kernel version is less than 4.17, the requested PVC sizes will not be enforced. Storage quotas will only be enforced on newer kernels.
Specific configurations for some distributions.
For NixOS, the kernel modules will be found in the non-standard path
/run/current-system/kernel-modules/lib/modules/, and they'll be symlinked inside the also non-standard path
Rook containers require read access to those locations to be able to load the required modules. They have to be bind-mounted as volumes in the CephFS and RBD plugin pods.
If installing Rook with Helm, uncomment these example settings in
If deploying without Helm, add those same values to the settings in the
rook-ceph-operator-config ConfigMap found in operator.yaml: