Skip to main content

Piraeus Operator Documentation

Deployment​

This project README contains documentation on the initial deployment process. A quick summary can be found here:

# Create an initial set of persistent volumes for etcd. Creates a hostpath volume on every control-plane node.
helm install piraeus-etcd-pv ./charts/pv-hostpath
# Deploy the piraeus operator chart. Replace <image> with the piraeus DRBD loader image matching your host OS.
helm install piraeus-op ./charts/piraeus --set operator.satelliteSet.kernelModuleInjectionImage=<image>

Then, wait for the deployment to finish:

kubectl wait  --for=condition=Ready --timeout=10m pod --all

The LINSTOR client​

Piraeus uses LINSTOR as the storage backend. Most configuration needs can be handled by the Piraeus Operator by editing one of the LinstorController, LinstorSatelliteSet or LinstorCSIDriver resources.

However, in some cases you might want to directly interface with the LINSTOR system using the linstor command. There are two ways to achieve this:

  • Use the kubectl-linstor plugin. The plugin interfaces with the resources used by the Piraeus Operator and enables you to use kubectl linstor ... to execute LINSTOR commands.
  • Execute the linstor command directly in the controller pod:
    kubectl exec -it deployment/piraeus-op-cs-controller -- linstor ...

Configuring storage pools​

To provision volumes, you need to configure storage pools. The LINSTOR backend supports a range of different storage providers. For some common providers, the Piraeus Operator provides convenient configuration via the LinstorSatelliteSet resource. You can read more on how to configure storage here.

Creating volumes​

Once you have storage pools configured (confirm by running kubectl linstor storage-pool list), you can almost start creating Persistent Volumes (PV) with Piraeus. First you will need to create a new storage class in Kubernetes.

The following example storage class configures piraeus to:

  • use 2 replicas for every persistent volume
  • use the xfs filesystem
  • use storage pools named ssd
  • allow volume expansion by resizing the Persistent Volume Claim (PVC).
  • wait for the first Pod to create the volume, placing a replica on the same node if possible.
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: piraeus-ssd
provisioner: linstor.csi.linbit.com
allowVolumeExpansion: true
volumeBindingMode: WaitForFirstConsumer
parameters:
  autoPlace: "2"
  storagePool: ssd
  csi.storage.k8s.io/fstype: xfs

You can find a full list of supported options here.

Using this storage class, you can provision volumes by applying a Persistent Volume Claim and waiting for Piraeus to provision the PV. The following PVC creates a 5GiB volume using the above storage class:

apiVersion: v1
  kind: PersistentVolumeClaim
  metadata:
    name: piraeus-pvc-1
  spec:
    storageClassName: piraeus-ssd
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 5Gi

Snapshots​

Piraeus supports snapshots via the CSI snapshotting feature. To enable this feature in your cluster, you need to add a Snapshot Controller to your cluster.

Some Kubernetes distributions (for example: OpenShift) already bundle this snapshot controller. On distributions without a bundled snapshot controller, you can use our guide here.

Additional information

Should you require further information check out the following links: