Skip to content

Configuring the storage cluster

Configure the storage cluster through the rook-local-cluster chart under systemConfiguration on the Cluster resource (or via the superphenix-operator Helm chart). The same pattern applies to both hyperconverged and decoupled deployments: set these values on the cluster that hosts Ceph.

The parameters below are the ones changed most often: Ceph network address ranges and which disks Rook claims for OSDs.

Where to apply this configuration

Configure storage on the cluster that hosts Ceph:

  • Hyperconverged: the single AZ cluster (storage and workloads together).
  • Decoupled: the storage cluster (not the workload cluster).
systemConfiguration:
  rook-local-cluster:
    helm:
      values:
        cephClusterSpec:
          network:
            addressRanges:
              public:
                - fd00:ffff:2000::/64
              cluster:
                # We reserve the first of the 65536 subnets to Ceph; nodes should only be
                # assigned addresses in that range.
                - fd00:ffff:2001::/96
          storage:
            deviceFilter: "nvme*|sd*"
  • network.addressRanges.public: address range for Ceph client (public) traffic.
  • network.addressRanges.cluster: address range for Ceph replication and recovery traffic. Keep node addresses within the reserved range.
  • storage.deviceFilter: which disks Rook claims as OSDs (for example NVMe and SCSI disks matching nvme* or sd*).

For network planning around public vs cluster storage traffic, see Network requirements. For disk sizing and Ceph media guidance, see Hardware requirements.

For advanced options, see the rook-ceph-cluster Helm chart values.