Chapter 4. Using Container Storage Interface (CSI)
4.1. Configuring CSI volumes
The Container Storage Interface (CSI) allows OpenShift Container Platform to consume storage from storage back ends that implement the CSI interface as persistent storage.
OpenShift Container Platform does not ship with any CSI drivers. It is recommended to use the CSI drivers provided by community or storage vendors.
Installation instructions differ by driver, and are found in each driver’s documentation. Follow the instructions provided by the CSI driver.
4.1.1. CSI Architecture
CSI drivers are typically shipped as container images. These containers are not aware of OpenShift Container Platform where they run. To use CSI-compatible storage back end in OpenShift Container Platform, the cluster administrator must deploy several components that serve as a bridge between OpenShift Container Platform and the storage driver.
The following diagram provides a high-level overview about the components running in pods in the OpenShift Container Platform cluster.
It is possible to run multiple CSI drivers for different storage back ends. Each driver needs its own external controllers deployment and daemon set with the driver and CSI registrar.
4.1.1.1. External CSI controllers
External CSI Controllers is a deployment that deploys one or more pods with three containers:
-
An external CSI attacher container translates
attach
anddetach
calls from OpenShift Container Platform to respectiveControllerPublish
andControllerUnpublish
calls to the CSI driver. -
An external CSI provisioner container that translates
provision
anddelete
calls from OpenShift Container Platform to respectiveCreateVolume
andDeleteVolume
calls to the CSI driver. - A CSI driver container
The CSI attacher and CSI provisioner containers communicate with the CSI driver container using UNIX Domain Sockets, ensuring that no CSI communication leaves the pod. The CSI driver is not accessible from outside of the pod.
attach
, detach
, provision
, and delete
operations typically require the CSI driver to use credentials to the storage backend. Run the CSI controller pods on infrastructure nodes so the credentials are never leaked to user processes, even in the event of a catastrophic security breach on a compute node.
The external attacher must also run for CSI drivers that do not support third-party attach
or detach
operations. The external attacher will not issue any ControllerPublish
or ControllerUnpublish
operations to the CSI driver. However, it still must run to implement the necessary OpenShift Container Platform attachment API.
4.1.1.2. CSI driver daemon set
The CSI driver daemon set runs a pod on every node that allows OpenShift Container Platform to mount storage provided by the CSI driver to the node and use it in user workloads (pods) as persistent volumes (PVs). The pod with the CSI driver installed contains the following containers:
-
A CSI driver registrar, which registers the CSI driver into the
openshift-node
service running on the node. Theopenshift-node
process running on the node then directly connects with the CSI driver using the UNIX Domain Socket available on the node. - A CSI driver.
The CSI driver deployed on the node should have as few credentials to the storage back end as possible. OpenShift Container Platform will only use the node plug-in set of CSI calls such as NodePublish
/NodeUnpublish
and NodeStage
/NodeUnstage
, if these calls are implemented.
4.1.2. Dynamic provisioning
Dynamic provisioning of persistent storage depends on the capabilities of the CSI driver and underlying storage back end. The provider of the CSI driver should document how to create a storage class in OpenShift Container Platform and the parameters available for configuration.
The created storage class can be configured to enable dynamic provisioning.
Procedure
Create a default storage class that ensures all PVCs that do not require any special storage class are provisioned by the installed CSI driver.
# oc create -f - << EOF apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: <storage-class> 1 annotations: storageclass.kubernetes.io/is-default-class: "true" provisioner: <provisioner-name> 2 parameters: EOF
4.1.3. Example using the CSI driver
The following example installs a default MySQL template without any changes to the template.
Prerequisites
- The CSI driver has been deployed.
- A storage class has been created for dynamic provisioning.
Procedure
Create the MySQL template:
# oc new-app mysql-persistent --> Deploying template "openshift/mysql-persistent" to project default ... # oc get pvc NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE mysql Bound kubernetes-dynamic-pv-3271ffcb4e1811e8 1Gi RWO cinder 3s
4.2. CSI volume snapshots
This document describes how to use volume snapshots with supported Container Storage Interface (CSI) drivers to help protect against data loss in OpenShift Container Platform. Familiarity with persistent volumes is suggested.
CSI volume snapshot is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.
For more information about the support scope of Red Hat Technology Preview features, see https://access.redhat.com/support/offerings/techpreview/.
4.2.1. Overview of CSI volume snapshots
A snapshot represents the state of the storage volume in a cluster at a particular point in time. Volume snapshots can be used to provision a new volume.
OpenShift Container Platform supports CSI volume snapshots by default. However, a specific CSI driver is required.
With CSI volume snapshots, a cluster administrator can:
- Deploy a third-party CSI driver that supports snapshots.
- Create a new persistent volume claim (PVC) from an existing volume snapshot.
- Take a snapshot of an existing PVC.
- Restore a snapshot as a different PVC.
- Delete an existing volume snapshot.
With CSI volume snapshots, an app developer can:
- Use volume snapshots as building blocks for developing application- or cluster-level storage backup solutions.
- Rapidly rollback to a previous development version.
- Use storage more efficiently by not having to make a full copy each time.
Be aware of the following when using volume snapshots:
- Support is only available for CSI drivers. In-tree and FlexVolumes are not supported.
- OpenShift Container Platform does not ship with any CSI drivers. It is recommended to use the CSI drivers provided by community or storage vendors. Follow the installation instructions provided by the CSI driver.
-
CSI drivers may or may not have implemented the volume snapshot functionality. CSI drivers that have provided support for volume snapshots will likely use the
csi-external-snapshotter
sidecar. See documentation provided by the CSI driver for details. - OpenShift Container Platform 4.4 supports version 1.1.0 of the CSI specification.
4.2.2. CSI snapshot controller and sidecar
OpenShift Container Platform provides a snapshot controller that is deployed into the control plane. In addition, your CSI driver vendor provides the CSI snapshot sidecar as a helper container that is installed during the CSI driver installation.
The CSI snapshot controller and sidecar provide volume snapshotting through the OpenShift Container Platform API. These external components run in the cluster.
The external controller is deployed by the CSI Snapshot Controller Operator.
4.2.2.1. External controller
The CSI snapshot controller binds VolumeSnapshot
and VolumeSnapshotContent
objects. The controller manages dynamic provisioning by creating and deleting VolumeSnapshotContent
objects.
4.2.2.2. External sidecar
Your CSI driver vendor provides the csi-external-snapshotter
sidecar. This is a separate helper container that is deployed with the CSI driver. The sidecar manages snapshots by triggering CreateSnapshot
and DeleteSnapshot
operations. Follow the installation instructions provided by your vendor.
4.2.3. About the CSI Snapshot Controller Operator
The CSI Snapshot Controller Operator runs in the openshift-cluster-storage-operator
namespace. It is installed by the Cluster Version Operator (CVO) in all clusters by default.
The CSI Snapshot Controller Operator installs the CSI snapshot controller, which runs in the csi-snapshot-controller
namespace.
4.2.3.1. Volume snapshot CRDs
During OpenShift Container Platform installation, the CSI Snapshot Controller Operator creates the following snapshot custom resource definitions (CRDs) in the snapshot.storage.k8s.io/
API group:
VolumeSnapshotContent
A snapshot taken of a volume in the cluster that has been provisioned by a cluster administrator.
Similar to the
PersistentVolume
CRD, theVolumeSnapshotContent
CRD is a cluster resource that points to a real snapshot in the storage back end.For manually pre-provisioned snapshots, a cluster administrator creates a number of
VolumeSnapshotContent
objects. These carry the details of the real volume snapshot in the storage system.The
VolumeSnapshotContent
CRD is not namespaced and is for use by a cluster administrator.VolumeSnapshot
Similar to the
PersistentVolumeClaim
CRD, theVolumeSnapshot
CRD defines a developer request for a snapshot. The CSI Snapshot Controller Operator runs the CSI snapshot controller, which handles the binding of aVolumeSnapshot
object with an appropriateVolumeSnapshotContent
object. The binding is a one-to-one mapping.The
VolumeSnapshot
CRD is namespaced. A developer uses the CRD as a distinct request for a snapshot.VolumeSnapshotClass
Allows a cluster administrator to specify different attributes belonging to a
VolumeSnapshot
object. These attributes may differ among snapshots taken of the same volume on the storage system, in which case they would not be expressed by using the same storage class of a persistent volume claim.The
VolumeSnapshotClass
CRD defines the parameters for thecsi-external-snapshotter
sidecar to use when creating a snapshot. This allows the storage back end to know what kind of snapshot to dynamically create if multiple options are supported.Dynamically provisioned snapshots use the
VolumeSnapshotClass
CRD to specify storage-provider-specific parameters to use when creating a snapshot.The
VolumeSnapshotContentClass
CRD is not namespaced and is for use by a cluster administrator to enable global configuration options for their storage back end.
4.2.4. Volume snapshot provisioning
There are two ways to provision snapshots: dynamically and manually.
4.2.4.1. Dynamic provisioning
Instead of using a preexisting snapshot, you can request that a snapshot be taken dynamically from a persistent volume claim. Parameters are specified using a VolumeSnapshotClass
CRD.
4.2.4.2. Manual provisioning
As a cluster administrator, you can manually pre-provision a number of VolumeSnapshotContent
objects. These carry the real volume snapshot details available to cluster users.
4.2.5. Creating a volume snapshot
When you create a VolumeSnapshot
object, OpenShift Container Platform creates a volume snapshot.
Prerequisites
- Logged in to a running OpenShift Container Platform cluster.
-
A PVC created using a CSI driver that supports
VolumeSnapshot
objects. - A storage class to provision the storage back end.
No pods are using the persistent volume claim (PVC) that you want to take a snapshot of.
NoteDo not create a volume snapshot of a PVC if a pod is using it. Doing so might cause data corruption because the PVC is not quiesced (paused). Be sure to first tear down a running pod to ensure consistent snapshots.
Procedure
To dynamically create a volume snapshot:
Create a file with the
VolumeSnapshotClass
object described by the following YAML:volumesnapshotclass.yaml
apiVersion: snapshot.storage.k8s.io kind: VolumeSnapshotClass 1 metadata: name: csi-hostpath-snap driver: hostpath.csi.k8s.io deletionPolicy: Delete
- 1
- Allows you to specify different attributes belonging to a volume snapshot.
Create the object you saved in the previous step by entering the following command:
$ oc create -f volumesnapshotclass.yaml
Create a
VolumeSnapshot
object:volumesnapshot-dynamic.yaml
apiVersion: snapshot.storage.k8s.io kind: VolumeSnapshot metadata: name: mysnap spec: volumeSnapshotClassName: csi-hostpath-snap 1 source: persistentVolumeClaimName: myclaim 2
- 1
- The request for a particular class by the volume snapshot. If
volumeSnapshotClassName
is empty, then no snapshot is created. - 2
- The name of the
PersistentVolumeClaim
object bound to a persistent volume. This defines what you want to create a snapshot of. Required for dynamically provisioning a snapshot.
Create the object you saved in the previous step by entering the following command:
$ oc create -f volumesnapshot-dynamic.yaml
To manually provision a snapshot:
Provide a value for the
volumeSnapshotContentName
parameter as the source for the snapshot, in addition to defining volume snapshot class as shown above.volumesnapshot-manual.yaml
apiVersion: snapshot.storage.k8s.io kind: VolumeSnapshot metadata: name: snapshot-demo spec: source: volumeSnapshotContentName: mycontent 1
- 1
- The
volumeSnapshotContentName
parameter is required for pre-provisioned snapshots.
Create the object you saved in the previous step by entering the following command:
$ oc create -f volumesnapshot-manual.yaml
Verification steps
After the snapshot has been created in the cluster, additional details about the snapshot are available.
To display details about the volume snapshot that was created, enter the following command:
$ oc describe volumesnapshot mysnap
The following example displays details about the
mysnap
volume snapshot:volumesnapshot.yaml
apiVersion: snapshot.storage.k8s.io kind: VolumeSnapshot metadata: name: mysnap spec: source: persistentVolumeClaimName: myclaim volumeSnapshotClassName: csi-hostpath-snap status: boundVolumeSnapshotContentName: snapcontent-1af4989e-a365-4286-96f8-d5dcd65d78d6 1 creationTime: "2020-01-29T12:24:30Z" 2 readyToUse: true 3 restoreSize: 500Mi
- 1
- The pointer to the actual storage content that was created by the controller.
- 2
- The time when the snapshot was created. The snapshot contains the volume content that was available at this indicated time.
- 3
- If the value is set to
true
, the snapshot can be used to restore as a new PVC.
If the value is set tofalse
, the snapshot was created. However, the storage back end needs to perform additional tasks to make the snapshot usable so that it can be restored as a new volume. For example, Amazon Elastic Block Store data might be moved to a different, less expensive location, which can take several minutes.
To verify that the volume snapshot was created, enter the following command:
$ oc get volumesnapshotcontent
The pointer to the actual content is displayed. If the
boundVolumeSnapshotContentName
field is populated, aVolumeSnapshotContent
object exists and the snapshot was created.-
To verify that the snapshot is ready, confirm that the
VolumeSnapshot
object hasreadyToUse: true
.
4.2.6. Deleting a volume snapshot
You can configure how OpenShift Container Platform deletes volume snapshots.
Procedure
To enable deletion of a volume snapshot in a cluster:
Specify the deletion policy that you require in the
VolumeSnapshotClass
object, as shown in the following example:volumesnapshot.yaml
apiVersion: snapshot.storage.k8s.io kind: VolumeSnapshotClass metadata: name: csi-hostpath-snap driver: hostpath.csi.k8s.io deletionPolicy: Delete 1
- 1
- If the
Delete
value is set, the underlying snapshot will be deleted, along with theVolumeSnapshotContent
object. If theRetain
value is set, both the underlying snapshot andVolumeSnapshotContent
object remain.
If theRetain
value is set, and theVolumeSnapshot
object is deleted without deleting the correspondingVolumeSnapshotContent
object, then the content will remain. The snapshot itself is also retained in the storage back end.
4.2.7. Restoring a volume snapshot
After your VolumeSnapshot
object is bound, you can use that object to provision a new volume that is pre-populated with data from the snapshot.
The volume snapshot content object is used to restore the existing volume to a previous state.
Prerequisites
- Logged in to a running OpenShift Container Platform cluster.
- A persistent volume claim (PVC) created using a Container Storage Interface (CSI) driver that supports volume snapshots.
- A storage class to provision the storage back end.
Procedure
Specify a
VolumeSnapshot
data source on a PVC as shown in the following:pvc-restore.yaml
apiVersion: v1 kind: PersistentVolumeClaim metadata: name: myclaim-restore spec: storageClassName: csi-hostpath-sc dataSource: name: mysnap 1 kind: VolumeSnapshot 2 apiGroup: snapshot.storage.k8s.io 3 accessModes: - ReadWriteOnce resources: requests: storage: 1Gi
Create a PVC by entering the following command:
$ oc create -f pvc-restore.yaml
Verify that the restored PVC has been created by entering the following command:
$ oc get pvc
Two different PVCs are displayed.
4.3. CSI volume cloning
Volume cloning duplicates an existing persistent volume to help protect against data loss in OpenShift Container Platform. This feature is only available with supported Container Storage Interface (CSI) drivers. You should be familiar with persistent volumes before you provision a CSI volume clone.
CSI volume cloning is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.
For more information about the support scope of Red Hat Technology Preview features, see https://access.redhat.com/support/offerings/techpreview/.
4.3.1. Overview of CSI volume cloning
A Container Storage Interface (CSI) volume clone is a duplicate of an existing persistent volume at a particular point in time.
Volume cloning is similar to volume snapshots, although it is more efficient. For example, a cluster administrator can duplicate a cluster volume by creating another instance of the existing cluster volume.
Cloning creates an exact duplicate of the specified volume on the back-end device, rather than creating a new empty volume. After dynamic provisioning, you can use a volume clone just as you would use any standard volume.
No new API objects are required for cloning. The existing dataSource
field in the PersistentVolumeClaim
object is expanded so that it can accept the name of an existing PersistentVolumeClaim in the same namespace.
4.3.1.1. Support limitations
By default, OpenShift Container Platform supports CSI volume cloning with these limitations:
- The destination persistent volume claim (PVC) must exist in the same namespace as the source PVC.
- The source and destination storage class must be the same.
- Support is only available for CSI drivers. In-tree and FlexVolumes are not supported.
- OpenShift Container Platform does not include any CSI drivers. Use the CSI drivers provided by community or storage vendors. Follow the installation instructions provided by the CSI driver.
- CSI drivers might not have implemented the volume cloning functionality. For details, see the CSI driver documentation.
- OpenShift Container Platform 4.4 supports version 1.1.0 of the CSI specification.
4.3.2. Provisioning a CSI volume clone
When you create a cloned persistent volume claim (PVC) API object, you trigger the provisioning of a CSI volume clone. The clone pre-populates with the contents of another PVC, adhering to the same rules as any other persistent volume. The one exception is that you must add a dataSource
that references an existing PVC in the same namespace.
Prerequisites
- You are logged in to a running OpenShift Container Platform cluster.
- Your PVC is created using a CSI driver that supports volume cloning.
- Your storage back end is configured for dynamic provisioning. Cloning support is not available for static provisioners.
Procedure
To clone a PVC from an existing PVC:
Create and save a file with the
PersistentVolumeClaim
object described by the following YAML:pvc-clone.yaml
apiVersion: v1 kind: PersistentVolumeClaim metadata: name: pvc-1-clone namespace: mynamespace spec: storageClassName: csi-cloning 1 accessModes: - ReadWriteOnce resources: requests: storage: 5Gi dataSource: kind: PersistentVolumeClaim name: pvc-1
- 1
- The name of the storage class that provisions the storage back end. The default storage class can be used and
storageClassName
can be omitted in the spec.
Create the object you saved in the previous step by running the following command:
$ oc create -f pvc-clone.yaml
A new PVC
pvc-1-clone
is created.Verify that the volume clone was created and is ready by running the following command:
$ oc get pvc pvc-1-clone
The
pvc-1-clone
shows that it isBound
.You are now ready to use the newly cloned PVC to configure a pod.
Create and save a file with the
Pod
object described by the YAML. For example:kind: Pod apiVersion: v1 metadata: name: mypod spec: containers: - name: myfrontend image: dockerfile/nginx volumeMounts: - mountPath: "/var/www/html" name: mypd volumes: - name: mypd persistentVolumeClaim: claimName: pvc-1-clone 1
- 1
- The cloned PVC created during the CSI volume cloning operation.
The created
Pod
object is now ready to consume, clone, snapshot, or delete your cloned PVC independently of its originaldataSource
PVC.