Verify that VMs stay up when disks are evicted

  • Related issues
    • #1334 Volumes fail with Scheduling Failure after evicting disc on multi-disc node
    • #5307 Replicas should be evicted and rescheduled to other disks before removing extrak disk

Verification Steps

  1. Created 3 nodes Harvester.

  2. Added formatted disk (called disk A) to node0 VM in the harvester node page.

  3. Added disk tag test on following disk in the longhorn page.

    1. disk A of node0
    2. root disk of node1
    3. root disk of node2

  4. Created storage class with disk tag test.

  5. Created volume (called B) with previous storage class. You should check scheduling status in the longhorn page.

  6. Created VM with volume B as extra volume (not boot volume).

  7. The node scheduling status should look like this. images/volumes/1334-image-01.png

  8. Deleted disk A in the harvester node page, error message should be displayed.

  9. Added disk tag test to root disk of node0.

  10. Requested eviction and disabled scheduling on disk A in the longhorn dashboard. Before: images/volumes/1334-image-02.png

    After: images/volumes/1334-image-03.png

  11. Removed disk A in the harvester node page, and it should be removed.

Expected Results

  1. Disk A is removed on the harvester node page.
  2. VM is running in whole steps.