Volumes are expanded in two stages. First, Longhorn expands the frontend (the block device), then it expands the filesystem.

To prevent the frontend expansion from interference by unexpected data R/W, Longhorn supports offline expansion only. The detached volume will be automatically attached to a random node with maintenance mode.

Rebuilding and adding replicas is not allowed during the expansion, and expansion is not allowed while replicas are rebuilding or being added.

If the volume was not expanded though the CSI interface (e.g. for Kubernetes older than v1.16), the capacity of the corresponding PVC and PV won’t change.

Prerequisite

  • The Longhorn version must be v0.8.0 or higher.
  • The volume to be expanded must be in the detached state.

Expand a Longhorn volume

There are two ways to expand a Longhorn volume: with a PersistentVolumeClaim (PVC) and with the Longhorn UI.

Via PVC

This method is applied only if:

  • The PVC is dynamically provisioned by the Kubernetes with Longhorn StorageClass.
  • The field allowVolumeExpansion should be true in the related StorageClass.

This method is recommended if it’s applicable, because the PVC and PV will be updated automatically and everything is kept consistent after expansion.

Usage: Find the corresponding PVC for Longhorn volume, then modify the requested spec.resources.requests.storage of the PVC:

  1. apiVersion: v1
  2. kind: PersistentVolumeClaim
  3. metadata:
  4. annotations:
  5. kubectl.kubernetes.io/last-applied-configuration: |
  6. {"apiVersion":"v1","kind":"PersistentVolumeClaim","metadata":{"annotations":{},"name":"longhorn-simple-pvc","namespace":"default"},"spec":{"accessModes":["ReadWriteOnce"],"resources":{"requests":{"storage":"1Gi"}},"storageClassName":"longhorn"}}
  7. pv.kubernetes.io/bind-completed: "yes"
  8. pv.kubernetes.io/bound-by-controller: "yes"
  9. volume.beta.kubernetes.io/storage-provisioner: driver.longhorn.io
  10. creationTimestamp: "2019-12-21T01:36:16Z"
  11. finalizers:
  12. - kubernetes.io/pvc-protection
  13. name: longhorn-simple-pvc
  14. namespace: default
  15. resourceVersion: "162431"
  16. selfLink: /api/v1/namespaces/default/persistentvolumeclaims/longhorn-simple-pvc
  17. uid: 0467ae73-22a5-4eba-803e-464cc0b9d975
  18. spec:
  19. accessModes:
  20. - ReadWriteOnce
  21. resources:
  22. requests:
  23. storage: 1Gi
  24. storageClassName: longhorn
  25. volumeMode: Filesystem
  26. volumeName: pvc-0467ae73-22a5-4eba-803e-464cc0b9d975
  27. status:
  28. accessModes:
  29. - ReadWriteOnce
  30. capacity:
  31. storage: 1Gi
  32. phase: Bound

Via Longhorn UI

Usage: On the volume page of Longhorn UI, click Expand for the volume.

Filesystem expansion

Longhorn will try to expand the file system only if:

  • The expanded size should be greater than the current size.
  • There is a Linux filesystem in the Longhorn volume.
  • The filesystem used in the Longhorn volume is one of the following:
    • ext4
    • XFS
  • The Longhorn volume is using the block device frontend.

Handling Volume Revert

If a volume is reverted to a snapshot with smaller size, the frontend of the volume is still holding the expanded size. But the filesystem size will be the same as that of the reverted snapshot. In this case, you will need to handle the filesystem manually:

  1. Attach the volume to a random node.

  2. Log in to the corresponding node, and expand the filesystem.

    If the filesystem is ext4, the volume might need to be mounted and umounted once before resizing the filesystem manually. Otherwise, executing resize2fs might result in an error:

    1. resize2fs: Superblock checksum does not match superblock while trying to open ......
    2. Couldn't find valid filesystem superblock.

    Follow the steps below to resize the filesystem:

    1. mount /dev/longhorn/<volume name> <arbitrary mount directory>
    2. umount /dev/longhorn/<volume name>
    3. mount /dev/longhorn/<volume name> <arbitrary mount directory>
    4. resize2fs /dev/longhorn/<volume name>
    5. umount /dev/longhorn/<volume name>
  3. If the filesystem is xfs, you can directly mount, then expand the filesystem.

    1. mount /dev/longhorn/<volume name> <arbitrary mount directory>
    2. xfs_growfs <the mount directory>
    3. umount /dev/longhorn/<volume name>