Storage class with single replica

You can create a storage class with a single replica to be used by your applications. This avoids redundant data copies and allows resiliency management on the application level.

About this task

Warning: Enabling this feature creates a single replica pool without data replication, increasing the risk of data loss, data corruption, and potential system instability if your application does not have its own replication. If any OSDs are lost, this feature requires very disruptive steps to recover. All applications can lose their data, and must be recreated in case of a failed OSD.

Procedure

  1. Enable the single replica feature using the following command: 
    $ oc patch storagecluster ocs-storagecluster -n openshift-storage --type json --patch '[{ "op": "replace", "path": "/spec/managedResources/cephNonResilientPools/enable", "value": true }]'
  2. Verify storagecluster is in Ready state: 
    $ oc get storagecluster
    Example output:
    
NAME                 AGE   PHASE   EXTERNAL   CREATED AT             VERSION
ocs-storagecluster   10m   Ready              2024-02-05T13:56:15Z   4.15.0
    New cephblockpools are created for each failure domain.
  3. Verify cephblockpools are in Ready state: 
    $ oc get cephblockpools
    Example output:
    
NAME                                          PHASE
ocs-storagecluster-cephblockpool              Ready
ocs-storagecluster-cephblockpool-us-east-1a   Ready
ocs-storagecluster-cephblockpool-us-east-1b   Ready
ocs-storagecluster-cephblockpool-us-east-1c   Ready
  4. Verify new storage classes have been created: 
    $ oc get storageclass
    Example output:
    
NAME                                        PROVISIONER                             RECLAIMPOLICY   VOLUMEBINDINGMODE      ALLOWVOLUMEEXPANSION   AGE
gp2 (default)                               kubernetes.io/aws-ebs                   Delete          WaitForFirstConsumer   true                   104m
gp2-csi                                     ebs.csi.aws.com                         Delete          WaitForFirstConsumer   true                   104m
gp3-csi                                     ebs.csi.aws.com                         Delete          WaitForFirstConsumer   true                   104m
ocs-storagecluster-ceph-non-resilient-rbd   openshift-storage.rbd.csi.ceph.com      Delete          WaitForFirstConsumer   true                   46m
ocs-storagecluster-ceph-rbd                 openshift-storage.rbd.csi.ceph.com      Delete          Immediate              true                   52m
ocs-storagecluster-cephfs                   openshift-storage.cephfs.csi.ceph.com   Delete          Immediate              true                   52m
openshift-storage.noobaa.io                 openshift-storage.noobaa.io/obc         Delete          Immediate              false                  50m
    New OSD pods are created; 3 osd-prepare pods and 3 additional pods.
  5. Verify new OSD pods are in Running state: 
    $ oc get pods | grep osd
    Example output:
    rook-ceph-osd-0-6dc76777bc-snhnm                                  2/2     Running     0               9m50s
rook-ceph-osd-1-768bdfdc4-h5n7k                                   2/2     Running     0               9m48s
rook-ceph-osd-2-69878645c4-bkdlq                                  2/2     Running     0               9m37s
rook-ceph-osd-3-64c44d7d76-zfxq9                                  2/2     Running     0               5m23s
rook-ceph-osd-4-654445b78f-nsgjb                                  2/2     Running     0               5m23s
rook-ceph-osd-5-5775949f57-vz6jp                                  2/2     Running     0               5m22s
rook-ceph-osd-prepare-ocs-deviceset-gp2-0-data-0x6t87-59swf       0/1     Completed   0               10m
rook-ceph-osd-prepare-ocs-deviceset-gp2-1-data-0klwr7-bk45t       0/1     Completed   0               10m
rook-ceph-osd-prepare-ocs-deviceset-gp2-2-data-0mk2cz-jx7zv       0/1     Completed   0               10m