Skip to main content
Version: 25.8

Delete a PersistentVolumeClaim

Follow the instructions on this page to delete PersistentVolumeClaims (PVCs) managed by PX-CSI. These operations help clean up unused resources.

Run the following command to remove a PVC that is no longer needed:

kubectl delete pvc <pvc-name> -n <pvc-namespace>
persistentvolumeclaim "<pvc-name>" deleted
important

Known limitation: For FlashArray file services, deleting a PVC doesn't delete the persistent volume (PV) in the FlashArray if the volume contains data and remains in the Released state. You must manually delete the PVs. For more information, see Delete persistent volumes in a released state in FA file services.

Delete persistent volumes in a released state in FA file services​

To delete PVs in the Released state, follow these steps:

  1. List the PVs in the Released state:

    kubectl get pv | grep Released
    pvc-xxxxxxxx-xxxx-xxxx-xxxx-105a73740493   50Gi       RWO   Delete   Released   fio/fio-log-fio-0   fa-file-sc-v4-authsys   <unset>   41m
    pvc-xxxxxxxx-xxxx-xxxx-xxxx-3aefabb5a065   200Gi      RWO   Delete   Released   fio/fio-data-fio-1  fa-file-sc-v4-authsys   <unset>   40m
    pvc-xxxxxxxx-xxxx-xxxx-xxxx-2b76f4288bdc   50Gi       RWO   Delete   Released   fio/fio-log-fio-1   fa-file-sc-v4-authsys   <unset>   40m
    pvc-xxxxxxxx-xxxx-xxxx-xxxx-ce081914311c   200Gi      RWO   Delete   Released   fio/fio-data-fio-0  fa-file-sc-v4-authsys   <unset>   41m

  2. Describe the PV to verify whether the deletion failed due to the volume containing data:

    kubectl describe pv <pvc-xxxxxxxx-xxxx-xxxx-xxxx-78be96424d27>
    Events:
    Type     Reason              Age                From                                                                                           Message
    ----     ------              ----               ----                                                                                           -------
    Warning  VolumeFailedDelete  83s (x3 over 86s)  pxd.portworx.com_px-pure-csi-controller-5d8d999569-h5trd_4e51323f-xxxx-xxxx-xxxx-964da5f4c298  persistentvolume pvc-xxxxxxxx-xxxx-xxxx-xxxx-78be96424d27 is still attached to node pwx-ocp-208-201-8rhmg-worker-0-dfh6t
    Warning  VolumeFailedDelete  23s (x4 over 79s)  pxd.portworx.com_px-pure-csi-controller-5d8d999569-h5trd_4e51323f-xxxx-xxxx-xxxx-964da5f4c298  rpc error: code = Internal desc = failed to delete backend volume: rpc error: code = Internal desc = delete failed: deletion of non-empty Directory not allowed

  3. Log in to a worker node and inspect the PVC mount to retrieve the volume label and NFS endpoint:

    mount |grep <pvc-xxxxxxxx-xxxx-xxxx-xxxx-78be96424d27>

  4. Mount the file system to a new directory so you can remove its contents:

    mkdir /var/lib/osd/mounts/pvc_xxxxxxxxxxx-clean-up
    mount -t nfs xx.xx.xxx.95:/px_20e4c7e1-pvc-xxxxxxxx-xxxx-xxxx-xxxx-78be96424d27 /var/lib/osd/mounts/1004978454102986342-clean-up

    In this example:

    • xx.xx.xxx.95 is the NFS endpoint.
    • px_20e4c7e1 consists of px and the prefix of the cluster UUID. For example, if the UUID is 20e4c7e1-xxxx-xxxx-xxxx-c367b2933167, use 20e4c7e1.

  5. Delete the directory contents and unmount the volume:

    rm -rf /var/lib/osd/mounts/1004978454102986342-clean-up/{..?*,.[!.]*,*}
    umount /var/lib/osd/mounts/1004978454102986342-clean-up
    rm -rf /var/lib/osd/mounts/1004978454102986342-clean-up

After completing these steps, the PV is automatically deleted.