Management vs. Workload (User) Clusters

We should concern ourselves only with Rancher Management Cluster(s) because Workload Clusters are deployed by Rancher Management Cluster (not HCC.)

The deployment of Workload Clusters is likely to be much less prescriptive and drive by common-sense Kubernetes-in-vSphere-VMs considerations.

Supportability

For NetApp HCI administrators HCC is the recommended and (perhaps more importantly) supported approach to deploy Rancher Management Cluster on NetApp HCI. For this deployment method, please refer to the NetApp documentation for Rancher on NetApp HCI.

ez-rancher can be used instead of HCC, but infrastructure-wise a Management Cluster deployed by ez-rancher should be indistinguishable from an HCC-deployed Management Cluster to ensure its supportability.

NetApp HCI users could deploy Management Cluster with HCC and then replicate (not literally - we don’t want IP address or other conflicts) with ez-rancher - this should be good enough for Dev/Test Management Clusters, for example. If we document those parameters by glancing at HCC deployment logs, we can then use ez-rancher to deploy Management Cluster in a manner consistent with HCC. Of course NetApp Support would be the judge of that, so the only reliable method to get a supported production Management Clusters is to deploy it with HCC.

vSphere administrators in a SolidFire environment have access to HCC, but HCC canonot be used to deploy Rancher. The reason is compute nodes where Rancher Management Cluster is deployed cannot be managed by HCC, so that task would be impossible to accomplish in the way it’s done on NetApp HCI compute nodes).SolidFire customers can therefore use ez-rancher or any method supported by Rancher (which includes manual installation of Virtual Machines). ez-rancher will be discussed in next section.

---