Applicable versions

Longhorn v1.4.x upgrading to v1.5.x

Symptoms

In some situations, volumes are unable to attach after Longhorn is upgraded from v1.4.x to v1.5.x. Workloads that use the affected volumes become stuck in the Pending state.

Possible Causes and Solutions

• Cause 1: Leftover non-empty volume.status.PendingNodeID

  This bug no longer exists in Longhorn v1.5.5 and later releases. If you encounter this bug in v1.5.0, v1.5.1, v1.5.2, v1.5.3, or v1.5.4, run the command kubectl edit volumes -n longhorn-system VOLUME-NAME --subresource=status and set volume.status.PendingNodeID to an empty string "".

• Cause 2: Unexpected longhorn-ui attachment ticket

  The root cause is still being investigated (see Issue #8339). If you encounter this issue, check the Longhorn VolumeAttachment CR by running the command kubectl -n longhorn-system edit volumeattachments.longhorn.io VOLUME-NAME. If the following longhorn-ui ticket exists, remove the whole block.

        longhorn-ui:
          generation: 0
          id: longhorn-ui
          nodeID: k3s-agent-acf067
          parameters:
            disableFrontend: "false"
          type: longhorn-api
  

  You can directly remove the whole block.

Related information

• https://github.com/longhorn/longhorn/issues/7994
• https://github.com/longhorn/longhorn/issues/8339