Degraded availability with added nodes
Volume creation using UI with degraded availability and added node
Prerequisites:
- Start with 1 node cluster.
- Double check if “Allow Volume Creation with Degraded Availability” is ticked or return true with following command:
kubectl get settings.longhorn.io/allow-volume-creation-with-degraded-availability -n longhorn-system
Steps:
- Create a Deployment Pod with a volume and three replicas.
- After the volume is attached, on Volume page it should be displayed as
Degraded
- Hover the cursor to the red circle exclamation mark, the tooltip will says, “The volume cannot be scheduled”.
- Click into the volume detail page it will display
Scheduling Failure
but the volume remain functional as expected.
- Write data to the Pod.
- Scale down the deployment to 0 to detach the volume.
- Volume return to
Detached
state.
- Both
Degraded
and Scheduling Failure
should be gone.
- Scale up the deployment back to 1 verify the data.
Scheduling Failure
should be seen again from the UI.
- Add another node to the cluster.
- Volume should start rebuilding on the second node soon.
- Once the rebuild completed, scale down and back up the deployment to verify the data.
- And the third node to the cluster.
- Volume should start rebuilding on the third node soon.
- Once the rebuilding starts, the
Scheduling Failure
should be gone.
- Scale down and back the deployment to verify the data.
[Edit]