Before performing maintenance, decommissioning hardware, or permanently removing a node from the cluster, it is recommended to first prevent new workloads from being assigned to that node and ensure any existing VODs have been removed.
Important Considerations
In VMS Gateway 9.2:
VODs are permanently assigned to the node on which they were created.
VODs are not automatically migrated to other nodes.
Removing a node without first removing its VODs will result in loss of service for those VODs.
Before taking a node out of service:
Identify any VODs running on the node.
Delete those VODs from the VMS Gateway UI.
Recreate the VODs on alternative nodes if required.
Preventing New Workloads from Being Scheduled
To mark a node as unavailable for future workloads, log on to the primary node and run:
sudo /opt/vms-gateway/bin/kubectl cordon <node-name>
You can verify the node status using:
sudo /opt/vms-gateway/bin/kubectl get nodes
A cordoned node remains operational but will not accept new workloads.
Removing the Node from the Cluster
Once all VODs have been removed and the node is no longer required:
# View all nodes
sudo /opt/vms-gateway/bin/kubectl get nodes
# Remove the node from the cluster
sudo /opt/vms-gateway/bin/kubectl delete node <node-name>
The node can now be removed from the VMS Gateway web interface.
After the node has been removed, it can be safely uninstalled or powered down.
Returning a Node to Service
If a node has only been cordoned and has not been removed from the cluster, it can be returned to service using:
sudo /opt/vms-gateway/bin/kubectl uncordon <node-name>
The node will once again be available to host new VODs.
Note: Since VMS Gateway 9.2 does not support automatic VOD migration or failover, cordoning a node is primarily useful for planned maintenance and preventing new workloads from being placed on that node. Existing VODs must still be manually recreated on another node if they need to be moved.