How to reconcile issues with Longhorn UI displaying 'Unknown' volume statuses or other inconsistencies
Various inconsistencies and errors may occur when using the Longhorn UI, specifically related to the display of volume statuses and snapshot listings. These anomalies could manifest as "unknown" volume statuses, incorrect or missing snapshot information, or other irregularities in the UI.
Root Cause: The instance managers in Longhorn are responsible for creating, monitoring, and managing block device instances (replicas and engines) on each node. Without direct management of the UI component, problems with instance managers could impact visibility in the UI and the ability to execute specific actions through the UI. Further details can be found in the 'Root Cause' section of this article.
Diagnosing:
- Volumes in the Longhorn UI are marked with an "Unknown" status.
- Snapshot listings appear incomplete, incorrect, or are missing altogether.
- The Longhorn UI may also present other inconsistent or erroneous data.
Below is an example where a PVC volume displays an "Unknown" status, accompanied by an error indicating that snapshot information could not be retrieved:
Solution: Delete the affected PVC's associated instance-manager pod.
- Copy the longhorn-instance-manager name from the 'Volume Details' pane after selecting the affected PVC within the Volume page
- In the example above, the value 'instance-manager-r-33620cae3d040fa11c56aaf7a6e631d2' from the would be copied from the Instance Manager field
- In the terminal, delete the associated instance-manager pod
- kubectl delete pod -n longhorn-system instance-manager-r-33620cae3d040fa11c56aaf7a6e631d2
- ensure to replace the 'instance-manager-r-33620cae3d040fa11c56aaf7a6e631d2' name with the name of your instance-manager associated with your affected PVC