Chapter 10. Managing container storage interface (CSI) component placements

Each cluster consists of a number of dedicated nodes such as infra and storage nodes. However, an infra node with a custom taint will not be able to use OpenShift Container Storage Persistent Volume Claims (PVCs) on the node. So, if you want to use such nodes, you can set tolerations to bring up csi-plugins on the nodes. For more information, see https://access.redhat.com/solutions/4827161.

Procedure

  1. Edit the configmap to add the toleration for the custom taint. Remember to save before exiting the editor. 

    $ oc edit configmap rook-ceph-operator-config -n openshift-storage

  2. Display the configmap to check the added toleration. 

    $ oc get configmap rook-ceph-operator-config -n openshift-storage -o yaml

    Example output of the added toleration for the taint, nodetype=infra:NoSchedule : 

    apiVersion: v1
data:
[...]
  CSI_PLUGIN_TOLERATIONS: |
    - key: nodetype
      operator: Equal
      value: infra
      effect: NoSchedule
    - key: node.ocs.openshift.io/storage
      operator: Equal
      value: "true"
      effect: NoSchedule
[...]
kind: ConfigMap
metadata:
[...]

  3. Restart the rook-ceph-operator if the csi-cephfsplugin-* and csi-rbdplugin-* pods fail to come up on their own on the infra nodes. 

    $ oc delete -n openshift-storage pod <name of the rook_ceph_operator pod>

    Example : 

    $ oc delete -n openshift-storage pod rook-ceph-operator-5446f9b95b-jrn2j

pod "rook-ceph-operator-5446f9b95b-jrn2j" deleted

Verification step

Verify that the csi-cephfsplugin-* and csi-rbdplugin-* pods are running on the infra nodes.