StorageOS Feature Labels

Feature labels are a powerful and flexible way to control storage features.

Applying specific feature labels triggers compression, replication and other storage features. No feature labels are present by default.

StorageOS Node labels

Nodes do not have any feature labels present by default. When StorageOS is run within Kubernetes, the StorageOS API Manager syncs any Kubernetes node labels to the corresponding StorageOS node. The Kubernetes node labels act as the “source of truth”, so labels should be applied to the Kubernetes nodes rather than to StorageOS nodes. This is because the Kubernetes node labels overwrite the StorageOS node labels on sync.

Feature	Label	Values	Description
Compute only	`storageos.com/computeonly`	true / false	Specifies whether a node should be `computeonly` where it only acts as a client and does not host volume data locally, otherwise the node is hyperconverged (the default), where the node can operate in both client and server modes.

You can set the computeonly label on the Kubernetes node and the label will be sync’d to the StorageOS node (labels take an eventual consistency reconciliation time of ~1min).

kubectl label node $NODE storageos.com/computeonly=true

StorageOS Volume labels

Volumes do not have any feature labels present by default.

WARNING: The encryption, caching and compression labels can only apply at provisioning time, they can’t be changed during execution.

Feature	Label	Values	Description
Caching	`storageos.com/nocache`	true / false	Switches off caching.
Compression	`storageos.com/nocompress`	true / false	Switches off compression of data at rest and in transit (compression is not enabled by default to maximise performance).
Encryption	`storageos.com/encryption`	true / false	Encrypts the contents of the volume. For each volume, a key is automatically generated, stored, and linked with the PVC.
Failure Mode	`storageos.com/failure-mode`	hard, soft, alwayson or integers [0, 5]	Sets the failure mode for a volume, either explicitly using a failure mode or implicitly using a replica threshold.
Replication	`storageos.com/replicas`	integers [0, 5]	Sets the number of replicas i.e full copies of the data across nodes. Typically 1 or 2 replicas is sufficient (2 or 3 instances of the data); latency implications need to be assesed when using more than 2 replicas.

To create a volume with a feature label:

Option 1: PVC Label

Add the label in the PVC definition, for instance:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: pvc-3
  labels:
    storageos.com/replicas: "1" # Label <-----
spec:
  storageClassName: "fast"
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 1G

Option 2: Set label in the StorageClass

Any PVC using the StorageClass inherits the label. The PVC label takes precedence over the StorageClass parameters.

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: storageos-replicated
parameters:
  csi.storage.k8s.io/fstype: ext4
  storageos.com/replicas: "1" # Label   <--------
provisioner: storageos # CSI driver (recommended)
# Change the NameSpace below if StorageOS doesn't run in kube-system
csi.storage.k8s.io/controller-expand-secret-name: csi-controller-expand-secret
csi.storage.k8s.io/controller-publish-secret-name: csi-controller-publish-secret
csi.storage.k8s.io/node-publish-secret-name: csi-node-publish-secret
csi.storage.k8s.io/provisioner-secret-name: csi-provisioner-secret
csi.storage.k8s.io/controller-expand-secret-namespace: kube-system   # NameSpace that runs StorageOS Daemonset
csi.storage.k8s.io/controller-publish-secret-namespace: kube-system  # NameSpace that runs StorageOS Daemonset
csi.storage.k8s.io/node-publish-secret-namespace: kube-system        # NameSpace that runs StorageOS Daemonset
csi.storage.k8s.io/provisioner-secret-namespace: kube-system         # NameSpace that runs StorageOS Daemonset

N.B. The StorageOS API manager periodically syncs labels from Kubernetes PVCs to the corresponding StorageOS volume. Therefore changes to StorageOS volume labels should be made to the corresponding Kubernetes PVC rather than to the StorageOS volume directly.

StorageOS Pod labels

Feature	Label	Values	Description
Pod fencing	`storageos.com/fenced`	true / false	Targets a pod to be fenced in case of node failure. (default: false)

For a pod to be fenced by StorageOS, a few requirements described in the Fencing Operations page need to be fulfilled.

kubectl label pod $POD storageos.com/fenced=true

It is recommended to define the fenced label in the pod’s manifest, i.e in the Statefulset definitions. Statefulsets pass labels to their VolumeClaimTemplates. You must set the label only at the spec.template.metadata.labels. Otherwise, the StorageOS volumes will fail to provision as only special accepted labels can be passed to volumes.

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: my-statefulset
spec:
  selector:
    matchLabels: # <----- Note that the matchLabels don't have the fenced label
      env: prod
  serviceName: my-statefulset-svc
  replicas: 1
  template:
    metadata:
      labels:   # <----- Note that the fenced label IS PRESENT
        env: prod
        storageos.com/fenced: "true"
    spec:
      containers:
...