Kubernetes and Trident objects

Both Kubernetes and Trident are designed to be interacted with through REST APIs by reading and writing resource objects.

Object overview

There are several different resource objects in play here, some that are managed through Kubernetes and others that are managed through Trident, that dictate the relationship between Kubernetes and Trident, Trident and storage, and Kubernetes and storage.

Perhaps the easiest way to understand these objects, what they are for and how they interact, is to follow a single request for storage from a Kubernetes user:

  1. A user creates a PersistentVolumeClaim requesting a new PersistentVolume of a particular size from a Kubernetes StorageClass that was previously configured by the administrator.
  2. The Kubernetes StorageClass identifies Trident as its provisioner and includes parameters that tell Trident how to provision a volume for the requested class.
  3. Trident looks at its own Trident StorageClass with the same name that identifies the matching Backends and StoragePools that it can use to provision volumes for the class.
  4. Trident provisions storage on a matching backend and creates two objects: a PersistentVolume in Kubernetes that tells Kubernetes how to find, mount and treat the volume, and a Volume in Trident that retains the relationship between the PersistentVolume and the actual storage.
  5. Kubernetes binds the PersistentVolumeClaim to the new PersistentVolume. Pods that include the PersistentVolumeClaim will mount that PersistentVolume on any host that it runs on.

Throughout the rest of this guide, we will describe the different Trident objects and details about how Trident crafts the Kubernetes PersistentVolume object for storage that it provisions.

For further reading about the Kubernetes objects, we highly recommend that you read the Persistent Volumes section of the Kubernetes documentation.

Kubernetes PersistentVolumeClaim objects

A Kubernetes PersistentVolumeClaim object is a request for storage made by a Kubernetes cluster user.

In addition to the standard specification, Trident allows users to specify the following volume-specific annotations if they want to override the defaults that you set in the backend configuration:

Annotation Volume Option Supported Drivers
trident.netapp.io/fileSystem fileSystem ontap-san, solidfire-san, eseries-iscsi
trident.netapp.io/cloneFromPVC cloneSourceVolume ontap-nas, ontap-san, solidfire-san, aws-cvs
trident.netapp.io/splitOnClone splitOnClone ontap-nas, ontap-san
trident.netapp.io/protocol protocol any
trident.netapp.io/exportPolicy exportPolicy ontap-nas, ontap-nas-economy, ontap-nas-flexgroup
trident.netapp.io/snapshotPolicy snapshotPolicy ontap-nas, ontap-nas-economy, ontap-nas-flexgroup, ontap-san
trident.netapp.io/snapshotReserve snapshotReserve ontap-nas, ontap-nas-flexgroup, ontap-san, aws-cvs
trident.netapp.io/snapshotDirectory snapshotDirectory ontap-nas, ontap-nas-economy, ontap-nas-flexgroup
trident.netapp.io/unixPermissions unixPermissions ontap-nas, ontap-nas-economy, ontap-nas-flexgroup
trident.netapp.io/blockSize blockSize solidfire-san

If the created PV has the Delete reclaim policy, Trident will delete both the PV and the backing volume when the PV becomes released (i.e., when the user deletes the PVC). Should the delete action fail, Trident will mark the PV as such and periodically retry the operation until it succeeds or the PV is manually deleted. If the PV uses the Retain policy, Trident ignores it and assumes the administrator will clean it up from Kubernetes and the backend, allowing the volume to be backed up or inspected before its removal. Note that deleting the PV will not cause Trident to delete the backing volume; it must be removed manually via the REST API (i.e., tridentctl).

One novel aspect of Trident is that users can provision new volumes by cloning existing volumes. Trident enables this functionality via the PVC annotation trident.netapp.io/cloneFromPVC. For example, if a user already has a PVC called mysql, she can create a new PVC called mysqlclone by referring to the mysql PVC: trident.netapp.io/cloneFromPVC: mysql. With this annotation set, Trident clones the volume corresponding to the mysql PVC, instead of provisioning a volume from scratch. A few points worth considering are the following: (1) We recommend cloning an idle volume, (2) a PVC and its clone must be in the same Kubernetes namespace and have the same storage class, and (3) with ontap-\* drivers, it might be desirable to set the PVC annotation trident.netapp.io/splitOnClone in conjunction with trident.netapp.io/cloneFromPVC. With trident.netapp.io/splitOnClone set to true, Trident splits the cloned volume from the parent volume; thus, completely decoupling the life cycle of the cloned volume from its parent at the expense of losing some storage efficiency. Not setting trident.netapp.io/splitOnClone or setting it to false results in reduced space consumption on the backend at the expense of creating dependencies between the parent and clone volumes such that the parent volume cannot be deleted unless the clone is deleted first. A scenario where splitting the clone makes sense is cloning an empty database volume where it’s expected for the volume and its clone to greatly diverge and not benefit from storage efficiencies offered by ONTAP.

sample-input/pvc-basic.yaml, sample-input/pvc-basic-clone.yaml, and sample-input/pvc-full.yaml contain examples of PVC definitions for use with Trident. See Trident Volume objects for a full description of the parameters and settings associated with Trident volumes.

Kubernetes PersistentVolume objects

A Kubernetes PersistentVolume object represents a piece of storage that’s been made available to the Kubernetes cluster. They have a lifecycle that’s independent of the pod that uses it.

Note

Trident creates PersistentVolume objects and registers them with the Kubernetes cluster automatically based on the volumes that it provisions. You are not expected to manage them yourself.

When a user creates a PVC that refers to a Trident-based StorageClass, Trident will provision a new volume using the corresponding storage class and register a new PV for that volume. In configuring the provisioned volume and corresponding PV, Trident follows the following rules:

  • Trident generates a PV name for Kubernetes and an internal name that it uses to provision the storage. In both cases it is assuring that the names are unique in their scope.
  • The size of the volume matches the requested size in the PVC as closely as possible, though it may be rounded up to the nearest allocatable quantity, depending on the platform.

Kubernetes StorageClass objects

Kubernetes StorageClass objects are specified by name in PersistentVolumeClaims to provision storage with a set of properties. The storage class itself identifies the provisioner that will be used and defines that set of properties in terms the provisioner understands.

It is one of two objects that need to be created and managed by you, the administrator. The other is the Trident Backend object.

A Kubernetes StorageClass object that uses Trident looks like this:

apiVersion: storage.k8s.io/v1beta1
kind: StorageClass
metadata:
  name: <Name>
provisioner: netapp.io/trident
mountOptions: <Mount Options>
parameters:
  <Trident Parameters>

These parameters are Trident-specific and tell Trident how to provision volumes for the class.

The storage class parameters are:

Attribute Type Required Description
attributes map[string]string no See the attributes section below
storagePools map[string]StringList no Map of backend names to lists of storage pools within
additionalStoragePools map[string]StringList no Map of backend names to lists of storage pools within
excludeStoragePools map[string]StringList no Map of backend names to lists of storage pools within

Storage attributes and their possible values can be classified into two groups:

  1. Storage pool selection attributes: These parameters determine which Trident-managed storage pools should be utilized to provision volumes of a given type.
Attribute Type Values Offer Request Supported by
media string hdd, hybrid, ssd Pool contains media of this type; hybrid means both Media type specified All drivers but aws-cvs
provisioningType string thin, thick Pool supports this provisioning method Provisioning method specified thick: all but solidfire-san & aws-cvs, thin: all but eseries-iscsi
backendType string
ontap-nas, ontap-nas-economy,
ontap-nas-flexgroup, ontap-san,
solidfire-san, eseries-iscsi, aws-cvs
Pool belongs to this type of backend Backend specified All drivers
snapshots bool true, false Pool supports volumes with snapshots Volume with snapshots enabled ontap-nas, ontap-san, solidfire-san, aws-cvs
clones bool true, false Pool supports cloning volumes Volume with clones enabled ontap-nas, ontap-san, solidfire-san, aws-cvs
encryption bool true, false Pool supports encrypted volumes Volume with encryption enabled ontap-nas, ontap-nas-economy, ontap-nas-flexgroups, ontap-san
IOPS int positive integer Pool is capable of guaranteeing IOPS in this range Volume guaranteed these IOPS solidfire-san

In most cases, the values requested will directly influence provisioning; for instance, requesting thick provisioning will result in a thickly provisioned volume. However, a SolidFire storage pool will use its offered IOPS minimum and maximum to set QoS values, rather than the requested value. In this case, the requested value is used only to select the storage pool.

Ideally you will be able to use attributes alone to model the qualities of the storage you need to satisfy the needs of a particular class. Trident will automatically discover and select storage pools that match all of the attributes that you specify.

If you find yourself unable to use attributes to automatically select the right pools for a class, you can use the storagePools and additionalStoragePools parameters to further refine the pools or even to select a specific set of pools manually.

The storagePools parameter is used to further restrict the set of pools that match any specified attributes. In other words, Trident will use the intersection of pools identified by the attributes and storagePools parameters for provisioning. You can use either parameter alone or both together.

The additionalStoragePools parameter is used to extend the set of pools that Trident will use for provisioning, regardless of any pools selected by the attributes and storagePools parameters.

The excludeStoragePools parameter is used to filter the set of pools that Trident will use for provisioning and will remove any pools that match.

In the storagePools and additionalStoragePools parameters, each entry takes the form <backend>:<storagePoolList>, where <storagePoolList> is a comma-separated list of storage pools for the specified backend. For example, a value for additionalStoragePools might look like ontapnas_192.168.1.100:aggr1,aggr2;solidfire_192.168.1.101:bronze. These lists will accept regex values for both the backend and list values. You can use tridentctl get backend to get the list of backends and their pools.

  1. Kubernetes attributes: These attributes have no impact on the selection of storage pools/backends by Trident during dynamic provisioning. Instead, these attributes simply supply parameters supported by Kubernetes Persistent Volumes.
Attribute Type Values Description Relevant Drivers Kubernetes Version
fsType string ext4, ext3, xfs, etc. The file system type for block volumes solidfire-san, ontap-san, eseries-iscsi All

The Trident installer bundle provides several example storage class definitions for use with Trident in sample-input/storage-class-*.yaml. Deleting a Kubernetes storage class will cause the corresponding Trident storage class to be deleted as well.

Trident StorageClass objects

Note

With Kubernetes, these objects are created automatically when a Kubernetes StorageClass that uses Trident as a provisioner is registered.

Trident creates matching storage classes for Kubernetes StorageClass objects that specify netapp.io/trident in their provisioner field. The storage class’s name will match that of the Kubernetes StorageClass object it represents.

Storage classes comprise a set of requirements for volumes. Trident matches these requirements with the attributes present in each storage pool; if they match, that storage pool is a valid target for provisioning volumes using that storage class.

One can create storage class configurations to directly define storage classes via the REST API. However, for Kubernetes deployments, we expect them to be created as a side-effect of registering new Kubernetes StorageClass objects.

Trident Backend objects

Backends represent the storage providers on top of which Trident provisions volumes; a single Trident instance can manage any number of backends.

This is one of the two object types that you will need to create and manage yourself. The other is the Kubernetes StorageClass object below.

For more information about how to construct these objects, visit the backend configuration guide.

Trident StoragePool objects

Storage pools represent the distinct locations available for provisioning on each backend. For ONTAP, these correspond to aggregates in SVMs. For SolidFire, these correspond to admin-specified QoS bands. For Cloud Volumes Service, these correspond to cloud provider regions. Each storage pool has a set of distinct storage attributes, which define its performance characteristics and data protection characteristics.

Unlike the other objects here, storage pool candidates are always discovered and managed automatically. View your backends to see the storage pools associated with them.

Trident Volume objects

Note

With Kubernetes, these objects are managed automatically and should not be manipulated by hand. You can view them to see what Trident provisioned, however.

Volumes are the basic unit of provisioning, comprising backend endpoints such as NFS shares and iSCSI LUNs. In Kubernetes, these correspond directly to PersistentVolumes. Each volume must be created with a storage class, which determines where that volume can be provisioned, along with a size.

A volume configuration defines the properties that a provisioned volume should have.

Attribute Type Required Description
version string no Version of the Trident API (“1”)
name string yes Name of volume to create
storageClass string yes Storage class to use when provisioning the volume
size string yes Size of the volume to provision in bytes
protocol string no Protocol type to use; “file” or “block”
internalName string no Name of the object on the storage system; generated by Trident
snapshotPolicy string no ontap-*: Snapshot policy to use
snapshotReserve string no ontap-*: Percentage of volume reserved for snapshots
exportPolicy string no ontap-nas*: Export policy to use
snapshotDirectory bool no ontap-nas*: Whether the snapshot directory is visible
unixPermissions string no ontap-nas*: Initial UNIX permissions
blockSize string no solidfire-*: Block/sector size
fileSystem string no File system type
cloneSourceVolume string no ontap-{nas|san} & solidfire-* & aws-cvs*: Name of the volume to clone from
splitOnClone string no ontap-{nas|san}: Split the clone from its parent

As mentioned, Trident generates internalName when creating the volume. This consists of two steps. First, it prepends the storage prefix – either the default, trident, or the prefix in the backend configurationd – to the volume name, resulting in a name of the form <prefix>-<volume-name>. It then proceeds to sanitize the name, replacing characters not permitted in the backend. For ONTAP backends, it replaces hyphens with underscores (thus, the internal name becomes <prefix>_<volume-name>), and for SolidFire, it replaces underscores with hyphens. For E-Series, which imposes a 30-character limit on all object names, Trident generates a random string for the internal name of each volume. For CVS (AWS), which imposes a 16-to-36-character limit on the unique volume creation token, Trident generates a random string for the internal name of each volume.

One can use volume configurations to directly provision volumes via the REST API, but in Kubernetes deployments we expect most users to use the standard Kubernetes PersistentVolumeClaim method. Trident will create this volume object automatically as part of the provisioning process in that case.