StorageClass
A StorageClass allows administrators to describe the classes of storage they offer. Different Longhorn StorageClasses might map to replica policies, or to node schedule policies, or disk schedule policies determined by the cluster administrators. This concept is sometimes called profiles in other storage systems.
note
For support with other storage, please refer to Third-Party Storage Support
Creating a StorageClass
- UI
- API
- Terraform
You can create one or more StorageClasses from the **Advanced > StorageClasses** page.
note
After a StorageClass is created, nothing can be changed except Description
.
Header Section
- Name: name of the StorageClass
- Description (optional): description of the StorageClass
Parameters Tab
Number of Replicas
The number of replicas created for each volume in Longhorn. Defaults to 3
.
Stale Replica Timeout
Determines when Longhorn would clean up an error replica after the replica’s status is ERROR. The unit is minute. Defaults to 30
minutes in Harvester.
Node Selector (Optional)
Select the node tags to be matched in the volume scheduling stage. You can add node tags by going to Host > Edit Config.
Disk Selector (Optional)
Select the disk tags to be matched in the volume scheduling stage. You can add disk tags by going to Host > Edit Config.
Migratable
Whether Live Migration is supported. Defaults to Yes
.
Customize Tab
Reclaim Policy
Volumes dynamically created by a StorageClass will have the reclaim policy specified in the reclaimPolicy
field of the class. The Delete
mode is used by default.
Delete
: Deletes volumes and the underlying devices when the volume claim is deleted.Retain
: Retains the volume for manual cleanup.
Allow Volume Expansion
Volumes can be configured to be expandable. This feature is Enabled
by default, which allows users to resize the volume by editing the corresponding PVC object.
note
You can only use the volume expansion feature to grow a Volume, not to shrink it.
Volume Binding Mode
The volumeBindingMode
field controls when volume binding and dynamic provisioning should occur. The Immediate
mode is used by default.
Immediate
: Binds and provisions a persistent volume once the PersistentVolumeClaim is created.WaitForFirstConsumer
: Binds and provisions a persistent volume once a VM using the PersistentVolumeClaim is created.
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
annotations:
storageclass.beta.kubernetes.io/is-default-class: 'true'
storageclass.kubernetes.io/is-default-class: 'true'
name: single-replica
parameters:
migratable: 'false'
numberOfReplicas: '1'
staleReplicaTimeout: '30'
provisioner: driver.longhorn.io
reclaimPolicy: Delete
volumeBindingMode: Immediate
allowVolumeExpansion: true
resource "harvester_storageclass" "single-replica" {
name = "single-replica"
is_default = "true"
allow_volume_expansion = "true"
volume_binding_mode = "immediate"
reclaim_policy = "delete"
parameters = {
"migratable" = "false"
"numberOfReplicas" = "1"
"staleReplicaTimeout" = "30"
}
}
Data Locality Settings
You can use the dataLocality
parameter when at least one replica of a Longhorn volume must be scheduled on the same node as the pod that uses the volume (whenever possible).
Harvester officially supports data locality as of v1.3.0. This applies even to volumes created from images. To configure data locality, create a new StorageClass on the Harvester UI (Storage Classess > Create > Parameters) and then add the following parameter:
- Key:
dataLocality
- Value:
disabled
orbest-effort
Data Locality Options
Harvester currently supports the following options:
disabled
: When applied, Longhorn may or may not schedule a replica on the same node as the pod that uses the volume. This is the default option.best-effort
: When applied, Longhorn always attempts to schedule a replica on the same node as the pod that uses the volume. Longhorn does not stop the volume even when a local replica is unavailable because of an environmental limitation (for example, insufficient disk space or incompatible disk tags).
note
Longhorn provides a third option called strict-local
, which forces Longhorn to keep only one replica on the same node as the pod that uses the volume. Harvester does not support this option because it can affect certain operations such as VM Live Migration
For more information, see Data Locality in the Longhorn documentation.
Appendix - Use Case
HDD Scenario
With the introduction of StorageClass, users can now use HDDs for tiered or archived cold storage.
caution
HDD is not recommended for guest RKE2 clusters or VMs with good performance disk requirements.
Recommended Practice
First, add your HDD on the Host
page and specify the disk tags as needed, such asHDD
or ColdStorage
. For more information on how to add extra disks and disk tags, see Multi-disk Management for details.
Then, create a new StorageClass
for the HDD (use the above disk tags). For hard drives with large capacity but slow performance, the number of replicas can be reduced to improve performance.
You can now create a volume using the above StorageClass
with HDDs mostly for cold storage or archiving purpose.