$ oc -n <hosted_cluster_namespace> scale nodepool <nodepool_name> \
--replicas 0
You can destroy a hosted cluster on x86
bare metal with IBM Z compute nodes and its managed cluster resource by using the command-line interface (CLI).
To destroy a hosted cluster and its managed cluster on x86
bare metal with IBM Z® compute nodes, you can use the command-line interface (CLI).
Scale the NodePool
object to 0
nodes by running the following command:
$ oc -n <hosted_cluster_namespace> scale nodepool <nodepool_name> \
--replicas 0
After the NodePool
object is scaled to 0
, the compute nodes are detached from the hosted cluster. In OKD version 4.17, this function is applicable only for IBM Z with KVM. For z/VM and LPAR, you must delete the compute nodes manually.
If you want to re-attach compute nodes to the cluster, you can scale up the NodePool
object with the number of compute nodes that you want. For z/VM and LPAR to reuse the agents, you must re-create them by using the Discovery
image.
If the compute nodes are not detached from the hosted cluster or are stuck in the
|
If you are using an OSA network device in Processor Resource/Systems Manager (PR/SM) mode, auto scaling is not supported. You must delete the old agent manually and scale up the node pool because the new agent joins during the scale down process.
Verify the status of the compute nodes by entering the following command:
$ oc --kubeconfig <hosted_cluster_name>.kubeconfig get nodes
After the compute nodes are detached from the hosted cluster, the status of the agents is changed to auto-assign
.
Delete the agents from the cluster by running the following command:
$ oc -n <hosted_control_plane_namespace> delete agent <agent_name>
You can delete the virtual machines that you created as agents after you delete the agents from the cluster. |
Destroy the hosted cluster by running the following command:
$ hcp destroy cluster agent --name <hosted_cluster_name> \
--namespace <hosted_cluster_namespace>